Bump to version 3.6.0

The agent will now only connect to Rooms when explicitly requested
via the Agent Dispatch API instead of auto-connecting to all new Rooms.

community/.env

@@ -30,9 +30,5 @@ MINIO_SECRET_KEY=minioadmin
 MONGO_ADMIN_USERNAME=mongoadmin
 MONGO_ADMIN_PASSWORD=mongoadmin
 
-# Default App (OpenVidu Call)
-CALL_PRIVATE_ACCESS=false
-CALL_USER=
-CALL_SECRET=
-CALL_ADMIN_USER=admin
-CALL_ADMIN_SECRET=admin
+# OpenVidu Meet base path
+MEET_BASE_PATH=/meet

community/agent-speech-processing.yaml

@@ -0,0 +1,437 @@
+# Docker image of the agent.
+docker_image: docker.io/openvidu/agent-speech-processing-vosk:3.6.0
+
+# Whether to run the agent or not.
+enabled: false
+
+# Maximum CPU load threshold for the agent to accept new jobs. Value between 0 and 1.
+load_threshold: 1.0
+
+# Log level for the agent [DEBUG, INFO, WARN, ERROR, CRITICAL]
+log_level: INFO
+
+live_captions:
+  # How this agent will connect to Rooms [manual, automatic]
+  # - manual: the agent will connect to new Rooms only when your application dictates it by using the Agent Dispatch API.
+  # - automatic: the agent will automatically connect to new Rooms.
+  processing: manual
+
+  # Which speech-to-text AI provider to use [aws, azure, google, openai, azure_openai, groq, deepgram, assemblyai, fal, clova, speechmatics, gladia, sarvam, mistralai, cartesia, soniox, nvidia, elevenlabs, simplismart, vosk, sherpa]
+  # The custom configuration for the selected provider must be set below
+  provider: vosk
+
+  aws:
+    # Credentials for AWS Transcribe. See https://docs.aws.amazon.com/transcribe/latest/dg/what-is.html
+    aws_access_key_id:
+    aws_secret_access_key:
+    aws_default_region:
+    # See https://docs.aws.amazon.com/transcribe/latest/dg/supported-languages.html
+    language:
+    # The name of the custom vocabulary you want to use.
+    # See https://docs.aws.amazon.com/transcribe/latest/dg/custom-vocabulary.html
+    vocabulary_name:
+    # The name of the custom language model you want to use.
+    # See https://docs.aws.amazon.com/transcribe/latest/dg/custom-language-models-using.html
+    language_model_name:
+    # Whether or not to enable partial result stabilization. Partial result stabilization can reduce latency in your output, but may impact accuracy.
+    # See https://docs.aws.amazon.com/transcribe/latest/dg/streaming-partial-results.html#streaming-partial-result-stabilization
+    enable_partial_results_stabilization:
+    # Specify the level of stability to use when you enable partial results stabilization (enable_partial_results_stabilization: true). Valid values: high | medium | low
+    # See https://docs.aws.amazon.com/transcribe/latest/dg/streaming-partial-results.html#streaming-partial-result-stabilization
+    partial_results_stability:
+    # The name of the custom vocabulary filter you want to use to mask or remove words.
+    # See https://docs.aws.amazon.com/transcribe/latest/dg/vocabulary-filtering.html
+    vocab_filter_name:
+    # The method used to filter the vocabulary. Valid values: mask | remove | tag
+    # See https://docs.aws.amazon.com/transcribe/latest/dg/vocabulary-filtering.html
+    vocab_filter_method:
+
+  azure:
+    # Credentials for Azure Speech Service.
+    # One of these combinations must be set:
+    #  - speech_host
+    #  - speech_key + speech_region
+    #  - speech_auth_token + speech_region
+    # See https://learn.microsoft.com/en-us/azure/ai-services/speech-service/get-started-speech-to-text?tabs=macos%2Cterminal&pivots=programming-language-python#prerequisites
+    speech_host:
+    speech_key:
+    speech_auth_token:
+    speech_region:
+    # Azure handles multiple languages and can auto-detect the language used. It requires the candidate set to be set. E.g. ["en-US", "es-ES"]
+    # See https://learn.microsoft.com/en-us/azure/ai-services/speech-service/language-support?tabs=stt#supported-languages
+    language:
+    # Removes profanity (swearing), or replaces letters of profane words with stars. Valid values: Masked | Removed | Raw
+    # See https://learn.microsoft.com/en-us/azure/ai-services/translator/profanity-filtering
+    profanity:
+    # List of words or phrases to boost recognition accuracy. Azure will give higher priority to these phrases during recognition.
+    phrase_list:
+    # Controls punctuation behavior. If True, enables explicit punctuation mode where punctuation marks are added explicitly. If False (default), uses Azure's default punctuation behavior.
+    explicit_punctuation:
+
+  azure_openai:
+    # Credentials for Azure OpenAI APIs. See https://learn.microsoft.com/en-us/azure/api-management/api-management-authenticate-authorize-azure-openai
+    # Azure OpenAI API key
+    azure_api_key:
+    # Azure Active Directory token
+    azure_ad_token:
+    # Azure OpenAI endpoint in the following format: https://{your-resource-name}.openai.azure.com. Mandatory value.
+    azure_endpoint:
+    # Name of your model deployment. If given with `azure_endpoint`, sets the base client URL to include `/deployments/{azure_deployment}`.
+    azure_deployment:
+    # OpenAI REST API version used for the request. Mandatory value.
+    api_version:
+    # OpenAI organization ID.
+    organization:
+    # OpenAI project ID.
+    project:
+    # The language code to use for transcription (e.g., "en" for English).
+    language:
+    # Whether to automatically detect the language.
+    detect_language:
+    # ID of the model to use for speech-to-text.
+    model:
+    # Initial prompt to guide the transcription.
+    prompt:
+
+  google:
+    # Credentials for Google Cloud. This is the content of a Google Cloud credential JSON file.
+    # Below is a dummy example for a credential type of "Service Account" (https://cloud.google.com/iam/docs/service-account-creds#key-types)
+    credentials_info: |
+      {
+        "type": "service_account",
+        "project_id": "my-project",
+        "private_key_id": "xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx",
+        "private_key": "-----BEGIN PRIVATE KEY-----\nxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx\n-----END PRIVATE KEY-----\n",
+        "client_email": "my-email@my-project.iam.gserviceaccount.com",
+        "client_id": "xxxxxxxxxxxxxxxxxxxxx",
+        "auth_uri": "https://accounts.google.com/o/oauth2/auth",
+        "token_uri": "https://oauth2.googleapis.com/token",
+        "auth_provider_x509_cert_url": "https://www.googleapis.com/oauth2/v1/certs",
+        "client_x509_cert_url": "https://www.googleapis.com/robot/v1/metadata/x509/my-email%40my-project.iam.gserviceaccount.com",
+        "universe_domain": "googleapis.com"
+      }
+    # Which model to use for recognition. If not set, uses the default model for the selected language.
+    # See https://cloud.google.com/speech-to-text/docs/transcription-model
+    model:
+    # The location to use for recognition. Default is "us-central1". Latency will be best if the location is close to your users.
+    # Check supported languages and locations at https://cloud.google.com/speech-to-text/v2/docs/speech-to-text-supported-languages
+    location:
+    # List of language codes to recognize. Default is ["en-US"].
+    # See https://cloud.google.com/speech-to-text/v2/docs/speech-to-text-supported-languages
+    languages:
+    # Whether to detect the language of the audio. Default is true.
+    detect_language:
+    # If 'true', adds punctuation to recognition result hypotheses. This feature is only available in select languages. Setting this
+    # for requests in other languages has no effect at all. The default 'false' value does not add punctuation to result hypotheses.
+    # See https://cloud.google.com/speech-to-text/docs/automatic-punctuation
+    punctuate:
+    # The spoken punctuation behavior for the call. If not set, uses default behavior based on model of choice.
+    # e.g. command_and_search will enable spoken punctuation by default. If 'true', replaces spoken punctuation
+    # with the corresponding symbols in the request. For example, "how are you question mark" becomes "how are you?".
+    # See https://cloud.google.com/speech-to-text/docs/spoken-punctuation for support. If 'false', spoken punctuation is not replaced.
+    spoken_punctuation:
+    # Whether to return interim (non-final) transcription results. Defaults to true.
+    interim_results:
+
+  openai:
+    # API key for OpenAI. See https://platform.openai.com/api-keys
+    api_key:
+    # The OpenAI model to use for transcription. See https://platform.openai.com/docs/guides/speech-to-text
+    model:
+    # The language of the input audio. Supplying the input language in ISO-639-1 format
+    # (https://en.wikipedia.org/wiki/List_of_ISO_639-1_codes) will improve accuracy and latency.
+    language:
+    # Whether to automatically detect the language.
+    detect_language:
+    # Optional text prompt to guide the transcription. Only supported for whisper-1.
+    prompt:
+
+  groq:
+    # API key for Groq. See https://console.groq.com/keys
+    api_key:
+    # See https://console.groq.com/docs/speech-to-text
+    model:
+    # The language of the input audio. Supplying the input language in ISO-639-1 format
+    # (https://en.wikipedia.org/wiki/List_of_ISO_639-1_codes) will improve accuracy and latency.
+    language:
+    # Whether to automatically detect the language.
+    detect_language:
+    # Prompt to guide the model's style or specify how to spell unfamiliar words. 224 tokens max.
+    prompt:
+    # Base URL for the Groq API. By default "https://api.groq.com/openai/v1"
+    base_url:
+
+  deepgram:
+    # See https://console.deepgram.com/
+    api_key:
+    # See https://developers.deepgram.com/reference/speech-to-text-api/listen-streaming#request.query.model
+    model:
+    # See https://developers.deepgram.com/reference/speech-to-text-api/listen-streaming#request.query.language
+    language:
+    # Whether to enable automatic language detection. See https://developers.deepgram.com/docs/language-detection
+    detect_language: false
+    # Whether to return interim (non-final) transcription results. See https://developers.deepgram.com/docs/interim-results
+    interim_results: true
+    # Whether to apply smart formatting to numbers, dates, etc. See https://developers.deepgram.com/docs/smart-format
+    smart_format: false
+    # When smart_format is used, ensures it does not wait for sequence to be complete before returning results. See https://developers.deepgram.com/docs/smart-format#using-no-delay
+    no_delay: true
+    # Whether to add punctuations to the transcription. Turn detector will work better with punctuations. See https://developers.deepgram.com/docs/punctuation
+    punctuate: true
+    # Whether to include filler words (um, uh, etc.) in transcription. See https://developers.deepgram.com/docs/filler-words
+    filler_words: true
+    # Whether to filter profanity from the transcription. See https://developers.deepgram.com/docs/profanity-filter
+    profanity_filter: false
+    # Whether to transcribe numbers as numerals. See https://developers.deepgram.com/docs/numerals
+    numerals: false
+    # List of tuples containing keywords and their boost values for improved recognition. Each tuple should be (keyword: str, boost: float). keywords does not work with Nova-3 models. Use keyterms instead.
+    # keywords:
+    #   - [OpenVidu, 1.5]
+    #   - [WebRTC, 1]
+    # List of key terms to improve recognition accuracy. keyterms is supported by Nova-3 models.
+    # Commented below is an example
+    keyterms:
+      # - "OpenVidu"
+      # - "WebRTC"
+
+  assemblyai:
+    # API key for AssemblyAI. See https://www.assemblyai.com/dashboard/api-keys
+    api_key:
+    # The confidence threshold (0.0 to 1.0) to use when determining if the end of a turn has been reached.
+    end_of_turn_confidence_threshold:
+    # The minimum amount of silence in milliseconds required to detect end of turn when confident.
+    min_end_of_turn_silence_when_confident:
+    # The maximum amount of silence in milliseconds allowed in a turn before end of turn is triggered.
+    max_turn_silence:
+    # Whether to return formatted final transcripts (proper punctuation, letter casing...). If enabled, formatted final transcripts are emitted shortly following an end-of-turn detection.
+    format_turns: true
+    # List of keyterms to improve recognition accuracy for specific words and phrases.
+    keyterms_prompt:
+      # - "OpenVidu"
+      # - "WebRTC"
+
+  fal:
+    # API key for fal. See https://fal.ai/dashboard/keys
+    api_key:
+    # See https://fal.ai/models/fal-ai/wizper/api#schema
+    language:
+
+  clova:
+    # Secret key issued when registering the app
+    api_key:
+    # API Gateway's unique invoke URL created in CLOVA Speech Domain.
+    # See https://guide.ncloud-docs.com/docs/en/clovaspeech-domain#create-domain
+    invoke_url:
+    # See https://api.ncloud-docs.com/docs/en/ai-application-service-clovaspeech-longsentence
+    language:
+    # Value between 0 and 1 indicating the threshold for the confidence score of the transcribed text. Default is 0.5.
+    # If the confidence score is lower than the threshold, the transcription event is not sent to the client.
+    # For a definition of the confidence score see https://api.ncloud-docs.com/docs/en/ai-application-service-clovaspeech-grpc
+    threshold:
+
+  speechmatics:
+    # API key for Speechmatics. See https://portal.speechmatics.com/manage-access/
+    api_key:
+    # ISO 639-1 language code. All languages are global and can understand different dialects/accents. To see the list of all supported languages, see https://docs.speechmatics.com/speech-to-text/languages#transcription-languages
+    language:
+    # Operating point to use for the transcription per required accuracy & complexity. To learn more, see https://docs.speechmatics.com/speech-to-text/languages#operating-points
+    operating_point:
+    # Partial transcripts allow you to receive preliminary transcriptions and update as more context is available until the higher-accuracy final transcript is returned. Partials are returned faster but without any post-processing such as formatting. See https://docs.speechmatics.com/speech-to-text/realtime/output#partial-transcripts
+    enable_partials:
+    # Enable speaker diarization. When enabled, the STT engine will determine and attribute words to unique speakers. The speaker_sensitivity parameter can be used to adjust the sensitivity of diarization
+    enable_diarization:
+    # RFC-5646 language code to make spelling rules more consistent in the transcription output. See https://docs.speechmatics.com/features/word-tagging#output-locale
+    output_locale:
+    # The delay in seconds between the end of a spoken word and returning the final transcript results. See https://docs.speechmatics.com/features/realtime-latency#configuration-example
+    max_delay:
+    # See https://docs.speechmatics.com/features/realtime-latency#configuration-example
+    max_delay_mode:
+    # Configuration for speaker diarization. See https://docs.speechmatics.com/features/diarization
+    speaker_diarization_config:
+      # See https://docs.speechmatics.com/features/diarization#max-speakers
+      max_speakers:
+      # See https://docs.speechmatics.com/features/diarization#speaker-sensitivity
+      speaker_sensitivity:
+      # See https://docs.speechmatics.com/features/diarization#prefer-current-speaker
+      prefer_current_speaker:
+    # Permitted punctuation marks for advanced punctuation. See https://docs.speechmatics.com/features/punctuation-settings
+    # Commented is an example of punctuation settings
+    punctuation_overrides:
+      # permitted_marks: [ ".", "," ]
+      # sensitivity: 0.4
+    # See https://docs.speechmatics.com/features/custom-dictionary
+    # Commented below is an example of a custom dictionary
+    additional_vocab:
+      # - content: financial crisis
+      # - content: gnocchi
+      #   sounds_like:
+      #     - nyohki
+      #     - nokey
+      #     - nochi
+      # - content: CEO
+      #   sounds_like:
+      #     - C.E.O.
+
+  gladia:
+    # API key for Gladia. See https://app.gladia.io/account
+    api_key:
+    # Whether to return interim (non-final) transcription results. Defaults to True
+    interim_results:
+    # List of language codes to use for recognition. Defaults to None (auto-detect). See https://docs.gladia.io/chapters/limits-and-specifications/languages
+    languages:
+    # Whether to allow switching between languages during recognition. Defaults to True
+    code_switching:
+    # https://docs.gladia.io/api-reference/v2/live/init#body-pre-processing-audio-enhancer
+    pre_processing_audio_enhancer:
+    # https://docs.gladia.io/api-reference/v2/live/init#body-pre-processing-speech-threshold
+    pre_processing_speech_threshold:
+
+  sarvam:
+    # API key for Sarvam. See https://dashboard.sarvam.ai/key-management
+    api_key:
+    # BCP-47 language code for supported Indian languages. See https://docs.sarvam.ai/api-reference-docs/speech-to-text/transcribe#request.body.language_code.language_code
+    language:
+    # The Sarvam STT model to use. See https://docs.sarvam.ai/api-reference-docs/speech-to-text/transcribe#request.body.model.model
+    model:
+
+  mistralai:
+    # API key for Mistral AI. See https://console.mistral.ai/api-keys
+    api_key:
+    # Name of the Voxtral STT model to use. Default to voxtral-mini-latest. See https://docs.mistral.ai/capabilities/audio/
+    model:
+    # The language code to use for transcription (e.g., "en" for English)
+    language:
+
+  cartesia:
+    # API key for Cartesia. See https://play.cartesia.ai/keys
+    api_key:
+    # The Cartesia STT model to use
+    model:
+    # The language code to use for transcription (e.g., "en" for English)
+    language:
+
+  soniox:
+    # API key for Soniox. See https://console.soniox.com/
+    api_key:
+    # Set language hints when possible to significantly improve accuracy. See: https://soniox.com/docs/stt/concepts/language-hints
+    language_hints:
+      # - "en"
+      # - "es"
+    # Set context to improve recognition of difficult and rare words. Context is a string and can include words, phrases, sentences, or summaries (limit: 10K chars). See https://soniox.com/docs/stt/concepts/context
+    context:
+
+  nvidia:
+    # API key for NVIDIA. See https://build.nvidia.com/explore/speech?integrate_nim=true&hosted_api=true&modal=integrate-nim
+    # Required when using NVIDIA's cloud services. To use a self-hosted NVIDIA Riva server setup "server" and "use_ssl" instead.
+    api_key:
+    # The NVIDIA Riva ASR model to use. Default is "parakeet-1.1b-en-US-asr-streaming-silero-vad-sortformer"
+    # See available models: https://build.nvidia.com/search/models?filters=usecase%3Ausecase_speech_to_text
+    model:
+    # The NVIDIA function ID for the model. Default is "1598d209-5e27-4d3c-8079-4751568b1081"
+    function_id:
+    # Whether to add punctuation to transcription results. Default is true.
+    punctuate:
+    # The language code for transcription. Default is "en-US"
+    language_code:
+    # Audio sample rate in Hz. Default is 16000.
+    sample_rate:
+    # The NVIDIA Riva server address. Default is "grpc.nvcf.nvidia.com:443"
+    # For self-hosted NIM, use your server address (e.g., "localhost:50051")
+    server:
+    # Whether to use SSL for the connection. Default is true.
+    # Set to false for locally hosted Riva NIM services without SSL.
+    use_ssl:
+
+  spitch:
+    # API key for Spitch. See https://docs.spitch.app/keys
+    api_key:
+    # Language short code for the generated speech. For supported values, see https://docs.spitch.app/concepts/languages
+    language:
+
+  elevenlabs:
+    # API key for ElevenLabs. See https://elevenlabs.io/app/settings/api-keys
+    api_key:
+    # The ElevenLabs STT model to use. Valid values are ["scribe_v1", "scribe_v2", "scribe_v2_realtime"]. See https://elevenlabs.io/docs/overview/models#models-overview
+    model_id:
+    # An ISO-639-1 or ISO-639-3 language_code corresponding to the language of the audio file. Can sometimes improve transcription performance if known beforehand. Defaults to null, in this case the language is predicted automatically
+    language_code:
+    # Custom base URL for the API. Optional.
+    base_url:
+    # Audio sample rate in Hz. Default is 16000.
+    sample_rate:
+    # Whether to tag audio events like (laughter), (footsteps), etc. in the transcription. Only supported for Scribe v1 model. Default is True
+    tag_audio_events:
+    # Whether to include word-level timestamps in the transcription. Default is false.
+    include_timestamps:
+
+  simplismart:
+    # API key for SimpliSmart. See https://docs.simplismart.ai/model-suite/settings/api-keys
+    api_key:
+    # Model identifier for the backend STT model. One of ["openai/whisper-large-v2", "openai/whisper-large-v3", "openai/whisper-large-v3-turbo"]
+    # Default is "openai/whisper-large-v3-turbo"
+    model:
+    # Language code for transcription (default: "en"). See https://docs.simplismart.ai/get-started/playground/transcription-models#supported-languages-with-their-codes
+    language:
+    # Operation to perform. "transcribe" converts speech to text in the original language, "translate" translates into English. Default is "transcribe".
+    task:
+    # If true, disables timestamp generation in transcripts. Default is true
+    without_timestamps:
+    # Minimum duration (ms) for a valid speech segment. Default is 0
+    min_speech_duration_ms:
+    # Decoding temperature (affects randomness). Default is 0.0
+    temperature:
+    # Whether to permit multilingual recognition. Default is false
+    multilingual:
+
+  vosk:
+    # Vosk language model. This provider requires docker_image "docker.io/openvidu/agent-speech-processing-vosk"
+    # Below is the list of pre-installed models in the container (available at https://alphacephei.com/vosk/models):
+    # - vosk-model-en-us-0.22-lgraph (English US)
+    # - vosk-model-small-cn-0.22 (Chinese)
+    # - vosk-model-small-de-0.15 (German)
+    # - vosk-model-small-en-in-0.4 (English India)
+    # - vosk-model-small-es-0.42 (Spanish)
+    # - vosk-model-small-fr-0.22 (French)
+    # - vosk-model-small-hi-0.22 (Hindi)
+    # - vosk-model-small-it-0.22 (Italian)
+    # - vosk-model-small-ja-0.22 (Japanese)
+    # - vosk-model-small-nl-0.22 (Dutch)
+    # - vosk-model-small-pt-0.3 (Portuguese)
+    # - vosk-model-small-ru-0.22 (Russian)
+    model: vosk-model-en-us-0.22-lgraph
+    # Language code for reference. It has no effect other than observability purposes.
+    # If a pre-installed "model" is declared, this will be set automatically if empty.
+    language:
+    # Audio sample rate in Hz. Default is 16000.
+    sample_rate:
+    # Whether to return interim/partial results during recognition. Default is true.
+    partial_results:
+    # Whether to override Vosk's built-in Voice Activity Detection (VAD) with Silero's VAD. Default is false.
+    use_silero_vad: false
+
+  sherpa:
+    # sherpa streaming model. This provider requires docker_image "docker.io/openvidu/agent-speech-processing-sherpa"
+    # Below is the list of pre-installed models in the container (available at https://github.com/k2-fsa/sherpa-onnx/releases/tag/asr-models):
+    # - sherpa-onnx-streaming-zipformer-en-kroko-2025-08-06 (English)
+    # - sherpa-onnx-streaming-zipformer-es-kroko-2025-08-06 (Spanish)
+    # - sherpa-onnx-streaming-zipformer-de-kroko-2025-08-06 (German)
+    # - sherpa-onnx-streaming-zipformer-fr-kroko-2025-08-06 (French)
+    # - sherpa-onnx-streaming-zipformer-ar_en_id_ja_ru_th_vi_zh-2025-02-10 (Multilingual: Arabic, English, Indonesian, Japanese, Russian, Thai, Vietnamese, Chinese)
+    model: sherpa-onnx-streaming-zipformer-en-kroko-2025-08-06
+    # Language code for reference. Auto-detected from model name if not set.
+    language:
+    # Runtime provider for sherpa-onnx. Supported values: "cpu" or "cuda". Default is "cpu".
+    # Learn about GPU acceleration at https://openvidu.io/docs/ai/live-captions/#gpu-acceleration-for-sherpa-provider
+    provider:
+    # Audio sample rate in Hz. Default is 16000.
+    sample_rate:
+    # Whether to return interim/partial results during recognition. Default is true.
+    partial_results:
+    # Number of threads for ONNX Runtime. Default is 2.
+    num_threads:
+    # Recognizer type ("transducer", "paraformer", "zipformer_ctc", "nemo_ctc", "t_one_ctc"). Auto-detected from model name if not set.
+    recognizer_type:
+    # Decoding method ("greedy_search", "modified_beam_search"). Default is "greedy_search".
+    decoding_method:
+    # Whether to override sherpa's built-in Voice Activity Detection (VAD) with Silero's VAD. Default is false.
+    use_silero_vad: false

community/configure_lan_private_ip_macos.sh

@@ -14,14 +14,4 @@ if [ -z "$LAN_PRIVATE_IP" ]; then
 fi
 
 # Replace the LAN_PRIVATE_IP in the .env file
-sed -i'' -e "s/LAN_PRIVATE_IP=.*/LAN_PRIVATE_IP=$LAN_PRIVATE_IP/g" .env
-
-# If sillicon mac, enable EXPERIMENTAL_DOCKER_DESKTOP_FORCE_QEMU flag
-if [ "$(uname -m)" = "arm64" ]; then
-    if ! grep -q "EXPERIMENTAL_DOCKER_DESKTOP_FORCE_QEMU" .env; then
-        echo "# Enable this flag to run Docker Desktop on Apple Silicon Macs" >> .env
-        echo "EXPERIMENTAL_DOCKER_DESKTOP_FORCE_QEMU=1" >> .env
-    else
-        sed -i'' -e "s/EXPERIMENTAL_DOCKER_DESKTOP_FORCE_QEMU=.*/EXPERIMENTAL_DOCKER_DESKTOP_FORCE_QEMU=1/g" .env
-    fi
-fi
+sed -i'' -e "s/LAN_PRIVATE_IP=.*/LAN_PRIVATE_IP=$LAN_PRIVATE_IP/g" .env

community/docker-compose.yaml

@@ -1,7 +1,6 @@
 services:
   caddy-proxy:
-    image: docker.io/openvidu/openvidu-caddy-local:3.1.0
-    platform: linux/amd64
+    image: docker.io/openvidu/openvidu-caddy-local:3.6.0
     container_name: caddy-proxy
     restart: unless-stopped
     extra_hosts:
@@ -17,37 +16,40 @@ services:
       - DASHBOARD_ADMIN_PASSWORD=${DASHBOARD_ADMIN_PASSWORD:-}
       - MINIO_ACCESS_KEY=${MINIO_ACCESS_KEY:-}
       - MINIO_SECRET_KEY=${MINIO_SECRET_KEY:-}
+      - MEET_BASE_PATH=${MEET_BASE_PATH:-/meet}
+    env_file:
+      - ./meet.env
     volumes:
       - ./custom-layout:/var/www/custom-layout
+      - /etc/localtime:/etc/localtime:ro
     ports:
       - 5443:5443
       - 6443:6443
       - 7443:7443
       - 7880:7880
+      - 9443:9443
+      - 9080:9080
     depends_on:
       setup:
         condition: service_completed_successfully
 
   redis:
-    image: docker.io/redis:7.4.2-alpine
-    platform: linux/amd64
+    image: docker.io/redis:8.6.1-alpine
     container_name: redis
     restart: unless-stopped
     ports:
       - 6379:6379
     volumes:
       - redis:/data
+      - /etc/localtime:/etc/localtime:ro
     command: >
-      redis-server
-      --bind 0.0.0.0
-      --requirepass ${REDIS_PASSWORD:-}
+      redis-server --bind 0.0.0.0 --requirepass ${REDIS_PASSWORD:-}
     depends_on:
       setup:
         condition: service_completed_successfully
 
   minio:
-    image: docker.io/bitnami/minio:2025.2.7-debian-12-r0
-    platform: linux/amd64
+    image: docker.io/openvidu/minio:2025.10.15-debian-12-r9
     container_name: minio
     restart: unless-stopped
     ports:
@@ -55,25 +57,27 @@ services:
     environment:
       - MINIO_ROOT_USER=${MINIO_ACCESS_KEY:-}
       - MINIO_ROOT_PASSWORD=${MINIO_SECRET_KEY:-}
-      - MINIO_DEFAULT_BUCKETS=openvidu
+      - MINIO_DEFAULT_BUCKETS=openvidu-appdata
       - MINIO_CONSOLE_SUBPATH=/minio-console
+      - MINIO_BROWSER=on
       - MINIO_BROWSER_REDIRECT_URL=http://localhost:7880/minio-console
     volumes:
       - minio-data:/bitnami/minio/data
       - minio-certs:/certs
+      - /etc/localtime:/etc/localtime:ro
     depends_on:
       setup:
         condition: service_completed_successfully
 
   mongo:
-    image: docker.io/bitnami/mongodb:8.0.4
-    platform: linux/amd64
+    image: docker.io/openvidu/mongodb:8.0.19-r1
     container_name: mongo
     restart: unless-stopped
     ports:
       - 27017:27017
     volumes:
       - mongo-data:/bitnami/mongodb
+      - /etc/localtime:/etc/localtime:ro
     environment:
       - MONGODB_ROOT_USER=${MONGO_ADMIN_USERNAME:-}
       - MONGODB_ROOT_PASSWORD=${MONGO_ADMIN_PASSWORD:-}
@@ -81,14 +85,12 @@ services:
       - MONGODB_REPLICA_SET_MODE=primary
       - MONGODB_REPLICA_SET_NAME=rs0
       - MONGODB_REPLICA_SET_KEY=devreplicasetkey
-      - EXPERIMENTAL_DOCKER_DESKTOP_FORCE_QEMU=${EXPERIMENTAL_DOCKER_DESKTOP_FORCE_QEMU:-0}
     depends_on:
       setup:
         condition: service_completed_successfully
 
   dashboard:
-    image: docker.io/openvidu/openvidu-dashboard:3.1.0
-    platform: linux/amd64
+    image: docker.io/openvidu/openvidu-dashboard:3.6.0
     container_name: dashboard
     restart: unless-stopped
     environment:
@@ -96,19 +98,21 @@ services:
       - ADMIN_USERNAME=${DASHBOARD_ADMIN_USERNAME:-}
       - ADMIN_PASSWORD=${DASHBOARD_ADMIN_PASSWORD:-}
       - DATABASE_URL=mongodb://${MONGO_ADMIN_USERNAME}:${MONGO_ADMIN_PASSWORD}@mongo:27017/?replicaSet=rs0&readPreference=primaryPreferred
+    volumes:
+      - /etc/localtime:/etc/localtime:ro
     depends_on:
       setup:
         condition: service_completed_successfully
 
   openvidu:
-    image: docker.io/openvidu/openvidu-server:3.1.0
-    platform: linux/amd64
+    image: docker.io/openvidu/openvidu-server:3.6.0
     restart: unless-stopped
     container_name: openvidu
     extra_hosts:
       - host.docker.internal:host-gateway
     environment:
       - LAN_PRIVATE_IP=${LAN_PRIVATE_IP:-}
+      - LAN_MODE=${LAN_MODE:-false}
     ports:
       - 3478:3478/udp
       - 7881:7881/tcp
@@ -118,15 +122,17 @@ services:
     volumes:
       - ./livekit.yaml:/etc/livekit.yaml
       - ./scripts/entrypoint_openvidu.sh:/scripts/entrypoint.sh
+      - /etc/localtime:/etc/localtime:ro
     depends_on:
       setup:
         condition: service_completed_successfully
 
   ingress:
-    image: docker.io/openvidu/ingress:3.1.0
-    platform: linux/amd64
+    image: docker.io/openvidu/ingress:3.6.0
     container_name: ingress
     restart: unless-stopped
+    extra_hosts:
+      - host.docker.internal:host-gateway
     ports:
       - 1935:1935
       - 8085:8085
@@ -135,13 +141,13 @@ services:
       - INGRESS_CONFIG_FILE=/etc/ingress.yaml
     volumes:
       - ./ingress.yaml:/etc/ingress.yaml
+      - /etc/localtime:/etc/localtime:ro
     depends_on:
       setup:
         condition: service_completed_successfully
 
   egress:
-    image: docker.io/livekit/egress:v1.9.0
-    platform: linux/amd64
+    image: docker.io/openvidu/egress:3.6.0
     restart: unless-stopped
     container_name: egress
     extra_hosts:
@@ -151,62 +157,85 @@ services:
     volumes:
       - ./egress.yaml:/etc/egress.yaml
       - egress-data:/home/egress/tmp
+      - /etc/localtime:/etc/localtime:ro
     depends_on:
       setup:
         condition: service_completed_successfully
 
-  default-app:
-    image: docker.io/openvidu/openvidu-call:3.1.0-demo
-    platform: linux/amd64
-    container_name: openvidu-call
+  openvidu-meet:
+    image: docker.io/openvidu/openvidu-meet:3.6.0
+    container_name: openvidu-meet
     restart: on-failure
+    extra_hosts:
+      - host.docker.internal:host-gateway
     environment:
       - USE_HTTPS=${USE_HTTPS:-false}
       - LAN_MODE=${LAN_MODE:-false}
       - LAN_DOMAIN=${LAN_DOMAIN:-}
       - LAN_PRIVATE_IP=${LAN_PRIVATE_IP:-}
-      - SERVER_PORT=6080
-      - CALL_NAME_ID=OpenViduCall-LOCAL
-      - LIVEKIT_URL_PRIVATE=ws://openvidu:7880/
       - LIVEKIT_API_KEY=${LIVEKIT_API_KEY}
       - LIVEKIT_API_SECRET=${LIVEKIT_API_SECRET}
-      - CALL_PRIVATE_ACCESS=${CALL_PRIVATE_ACCESS:-false}
-      - CALL_USER=${CALL_USER:-}
-      - CALL_SECRET=${CALL_SECRET:-}
-      - CALL_RECORDING=${CALL_RECORDING:-}
-      - CALL_ADMIN_USER=${CALL_ADMIN_USER:-admin}
-      - CALL_ADMIN_SECRET=${CALL_ADMIN_SECRET:-admin}
-      - CALL_LOG_LEVEL=${CALL_LOG_LEVEL:-info}
-      - CALL_S3_BUCKET=${CALL_S3_BUCKET:-openvidu}
-      - CALL_S3_SERVICE_ENDPOINT=${CALL_S3_SERVICE_ENDPOINT:-http://minio:9000}
-      - CALL_S3_ACCESS_KEY=${MINIO_ACCESS_KEY}
-      - CALL_S3_SECRET_KEY=${MINIO_SECRET_KEY}
-      - CALL_AWS_REGION=${CALL_AWS_REGION:-us-east-1}
-      - CALL_S3_WITH_PATH_STYLE_ACCESS=${CALL_S3_WITH_PATH_STYLE_ACCESS:-true}
+      - MEET_S3_ACCESS_KEY=${MINIO_ACCESS_KEY}
+      - MEET_S3_SECRET_KEY=${MINIO_SECRET_KEY}
+      - MEET_REDIS_PASSWORD=${REDIS_PASSWORD:-}
+      - MEET_MONGO_URI=mongodb://${MONGO_ADMIN_USERNAME}:${MONGO_ADMIN_PASSWORD}@mongo:27017/?replicaSet=rs0&readPreference=primaryPreferred
+      - MEET_BASE_PATH=${MEET_BASE_PATH:-/meet}
+      - MEET_CONFIG_DIR=/config/meet.env
     volumes:
-      - ./scripts/entrypoint_default_app.sh:/scripts/entrypoint.sh
+      - ./meet.env:/config/meet.env
+      - ./scripts/entrypoint_openvidu_meet.sh:/scripts/entrypoint.sh
       - ./scripts/utils.sh:/scripts/utils.sh
+      - /etc/localtime:/etc/localtime:ro
     entrypoint: /bin/sh /scripts/entrypoint.sh
     depends_on:
       setup:
         condition: service_completed_successfully
 
+  operator:
+    image: docker.io/openvidu/openvidu-operator:3.6.0
+    container_name: operator
+    restart: unless-stopped
+    volumes:
+      - /var/run/docker.sock:/var/run/docker.sock
+      - agents-config:/agents-config
+      - ./:/deployment
+      - /etc/localtime:/etc/localtime:ro
+    environment:
+      - MODE=agent-manager-local
+      - DEPLOYMENT_FILES_DIR=/deployment
+      - AGENTS_CONFIG_DIR=/agents-config
+      - NETWORK_NAME=openvidu-community
+      - AGENTS_CONFIG_VOLUME=openvidu-agents-config
+      - LIVEKIT_URL=ws://openvidu:7880/
+      - LIVEKIT_API_KEY=${LIVEKIT_API_KEY:-}
+      - LIVEKIT_API_SECRET=${LIVEKIT_API_SECRET:-}
+      - REDIS_ADDRESS=redis:6379
+      - REDIS_PASSWORD=${REDIS_PASSWORD:-}
+    depends_on:
+      setup:
+        condition: service_completed_successfully
+
   ready-check:
-    image: docker.io/curlimages/curl:8.12.0
-    platform: linux/amd64
+    image: docker.io/openvidu/openvidu-operator:3.6.0
     container_name: ready-check
     restart: on-failure
+    volumes:
+      - /etc/localtime:/etc/localtime:ro
     environment:
+      - MODE=local-ready-check
+      - OPENVIDU_ENVIRONMENT=local-platform
       - USE_HTTPS=${USE_HTTPS:-false}
       - LAN_DOMAIN=${LAN_DOMAIN:-}
       - LAN_MODE=${LAN_MODE:-false}
       - LAN_PRIVATE_IP=${LAN_PRIVATE_IP:-}
-      - LIVEKIT_API_KEY=${LIVEKIT_API_KEY:-}
-      - LIVEKIT_API_SECRET=${LIVEKIT_API_SECRET:-}
       - DASHBOARD_ADMIN_USERNAME=${DASHBOARD_ADMIN_USERNAME:-}
       - DASHBOARD_ADMIN_PASSWORD=${DASHBOARD_ADMIN_PASSWORD:-}
       - MINIO_ACCESS_KEY=${MINIO_ACCESS_KEY:-}
       - MINIO_SECRET_KEY=${MINIO_SECRET_KEY:-}
+      - LIVEKIT_API_KEY=${LIVEKIT_API_KEY:-}
+      - LIVEKIT_API_SECRET=${LIVEKIT_API_SECRET:-}
+    env_file:
+      - ./meet.env
     depends_on:
       - openvidu
       - ingress
@@ -214,14 +243,9 @@ services:
       - dashboard
       - minio
       - mongo
-    volumes:
-      - ./scripts/ready-check.sh:/scripts/ready-check.sh
-      - ./scripts/utils.sh:/scripts/utils.sh
-    command: /bin/sh /scripts/ready-check.sh
 
   setup:
     image: docker.io/busybox:1.37.0
-    platform: linux/amd64
     container_name: setup
     restart: "no"
     volumes:
@@ -229,6 +253,7 @@ services:
       - mongo-data:/mongo
       - egress-data:/egress
       - ./scripts/setup.sh:/scripts/setup.sh
+      - /etc/localtime:/etc/localtime:ro
     environment:
       - USE_HTTPS=${USE_HTTPS:-false}
       - LAN_MODE=${LAN_MODE:-false}
@@ -238,9 +263,21 @@ services:
     command: /bin/sh /scripts/setup.sh
 
 volumes:
+  agents-config:
+    name: openvidu-agents-config
   minio-certs:
+    name: openvidu-minio-certs
   mongodb-config:
+    name: openvidu-mongodb-config
   redis:
+    name: openvidu-redis
   minio-data:
+    name: openvidu-minio-data
   mongo-data:
+    name: openvidu-mongo-data
   egress-data:
+    name: openvidu-egress-data
+
+networks:
+  default:
+    name: openvidu-community

community/egress.yaml

@@ -21,12 +21,18 @@ storage:
         # Default region for minio
         region: us-east-1
         endpoint: http://minio:9000
-        bucket: openvidu
+        bucket: openvidu-appdata
         force_path_style: true
+    #azure:
+    #    account_name: your_account_name
+    #    account_key: your_account_key
+    #    container_name: openvidu-appdata
+    # gcp:
+    #     credentials_json: |
+    #         your_credentials_json
+    #     bucket: openvidu-appdata
 
 # CPU cost for each type of Egress operation.
-# Defaults are set for unlimited operations.
-# Change these values according to your needs.
 cpu_cost:
     max_cpu_utilization: 0.80
     room_composite_cpu_cost: 0.01
@@ -36,3 +42,20 @@ cpu_cost:
     participant_cpu_cost: 0.01
     track_composite_cpu_cost: 0.01
     track_cpu_cost: 0.01
+
+openvidu:
+    # Allocation strategy for new egress requests
+    # - cpuload: the node with the lowest CPU load will be selected. Distributes the CPU load evenly across all nodes.
+    # - binpack: some node already hosting at least one egress will be selected. Fills up nodes before assigning work to new ones.
+    allocation_strategy: cpuload
+    # Whether to use system-wide CPU monitoring or egress process CPU monitoring. This affects the allocation of new egress requests.
+    # It is preferable to set this value to:
+    # - true: when the egress service is running in a shared server also hosting other CPU-intensive services.
+    # - false: when the egress service is running in a dedicated server.
+    use_global_cpu_monitoring: true
+    # Disables the automatic killing of the most expensive egress when CPU is overloaded.
+    # The default "false" value helps keeping the node stable, but may cause unexpected egress terminations under high load.
+    disable_cpu_overload_killer: false
+    # Minimum available disk space in MB required to accept new egress requests.
+    # Default: 512 MB. Set to a negative value (e.g., -1) to disable disk space checking.
+    min_disk_space_mb: 512

community/livekit.yaml

@@ -31,7 +31,7 @@ webhook:
     api_key: devkey
     urls:
         - http://host.docker.internal:6080/livekit/webhook
-        - http://default-app:6080/livekit/webhook
+        - http://openvidu-meet:6080/livekit/webhook
 ingress:
     rtmp_base_url: rtmp://localhost:1935/rtmp
     whip_base_url: http://localhost:8085/whip

community/meet.env

@@ -0,0 +1,36 @@
+# OpenVidu Meet configuration
+# Static environment variables loaded via MEET_CONFIG_DIR
+
+SERVER_PORT=6080
+MEET_NAME_ID=openviduMeet-LOCAL
+MEET_LOG_LEVEL=info
+MEET_COOKIE_SECURE=false
+MEET_INITIAL_ADMIN_USER=admin
+MEET_INITIAL_ADMIN_PASSWORD=admin
+MEET_INITIAL_API_KEY=meet-api-key
+MEET_INITIAL_WEBHOOK_ENABLED=true
+MEET_INITIAL_WEBHOOK_URL=http://host.docker.internal:6080/webhook
+
+LIVEKIT_URL_PRIVATE=ws://openvidu:7880/
+
+# S3 configuration
+MEET_S3_BUCKET=openvidu-appdata
+MEET_S3_SUBBUCKET=openvidu-meet
+MEET_S3_SERVICE_ENDPOINT=http://minio:9000
+MEET_AWS_REGION=us-east-1
+MEET_S3_WITH_PATH_STYLE_ACCESS=true
+
+# Storage backend type
+MEET_BLOB_STORAGE_MODE=s3
+
+# Redis configuration
+MEET_REDIS_HOST=redis
+MEET_REDIS_PORT=6379
+MEET_REDIS_DB=0
+
+# MongoDB configuration
+MEET_MONGO_ENABLED=true
+MEET_MONGO_DB_NAME=openvidu-meet
+
+# Enable live captions using OpenVidu Speech to Text agent
+MEET_CAPTIONS_ENABLED=false

community/scripts/entrypoint_openvidu.sh

@@ -1,7 +1,8 @@
 #!/bin/sh
 set -e
 
-if [ "$LAN_PRIVATE_IP" != "none" ]; then
+if [ "$LAN_PRIVATE_IP" != "" ] && [ "$LAN_MODE" = 'true' ]; then
+    echo "Using as NODE_IP: $LAN_PRIVATE_IP"
     export NODE_IP="$LAN_PRIVATE_IP"
 fi
 

community/scripts/entrypoint_openvidu_meet.sh

@@ -2,7 +2,7 @@
 
 . /scripts/utils.sh
 
-URL=$(getDeploymentUrl)
+URL=$(getDeploymentUrl ws)
 export LIVEKIT_URL="${URL}"
 
 /usr/local/bin/entrypoint.sh

community/scripts/ready-check.sh

@@ -1,70 +0,0 @@
-#!/bin/sh
-
-. /scripts/utils.sh
-
-trap 'handle_sigint' SIGINT
-
-handle_sigint() {
-  echo "SIGINT signal received, exiting..."
-  exit 1
-}
-
-wait_for_service() {
-  SERVICE_NAME=$1
-  SERVICE_URL=$2
-  shift 2
-  EXTRA=$@
-  if [ -n "$EXTRA" ]; then
-    until curl $EXTRA $SERVICE_URL > /dev/null; do
-      echo "Waiting for $SERVICE_NAME to start...";
-      sleep 1;
-    done;
-  else
-    until curl --silent --head --fail $SERVICE_URL > /dev/null; do
-      echo "Waiting for $SERVICE_NAME to start...";
-      sleep 1;
-    done;
-  fi;
-}
-
-wait_for_service 'OpenVidu' 'http://openvidu:7880'
-wait_for_service 'Ingress' 'http://ingress:9091'
-wait_for_service 'Egress' 'http://egress:9091'
-wait_for_service 'Dashboard' 'http://dashboard:5000'
-wait_for_service 'Minio' 'http://minio:9000/minio/health/live'
-wait_for_service 'Minio Console' 'http://minio:9001/minio-console'
-wait_for_service 'Mongo' 'http://mongo:27017' --connect-timeout 10 --silent
-
-LAN_HTTP_URL=$(getDeploymentUrl http)
-LAN_WS_URL=$(getDeploymentUrl ws)
-
-for i in $(seq 1 10); do
-  echo 'Starting OpenVidu... Please be patient...'
-  sleep 1
-done;
-echo ''
-echo ''
-echo '========================================='
-echo '🎉 OpenVidu is ready! 🎉'
-echo '========================================='
-echo ''
-echo 'OpenVidu Server & LiveKit Server URLs:'
-echo ''
-echo '    - From this machine:'
-echo ''
-echo '        - http://localhost:7880'
-echo '        - ws://localhost:7880'
-echo ''
-echo '    - From other devices in your LAN:'
-echo ''
-echo "        - $LAN_HTTP_URL"
-echo "        - $LAN_WS_URL"
-echo ''
-echo '========================================='
-echo ''
-echo 'OpenVidu Developer UI (services and passwords):'
-echo ''
-echo '    - http://localhost:7880'
-echo "    - $LAN_HTTP_URL"
-echo ''
-echo '========================================='

community/scripts/setup.sh

@@ -1,6 +1,11 @@
 #!/bin/sh
 
-if [ -z "$LAN_PRIVATE_IP" ]; then
+if [ "$LAN_MODE" = 'true' ] && [ "$USE_HTTPS" = 'false' ]; then
+  echo 'LAN_MODE cannot be "true" if USE_HTTPS is "false"'
+  exit 1
+fi
+
+if [ "$LAN_MODE" = 'true' ] && [ -z "$LAN_PRIVATE_IP" ]; then
   echo '------------------------'
   echo ''
   echo 'LAN_PRIVATE_IP is required in the .env file.'
@@ -14,7 +19,9 @@ if [ -z "$LAN_PRIVATE_IP" ]; then
   echo 'If it can'\''t be found, you can manually set it in the .env file'
   echo '------------------------'
   exit 1
-else
+fi
+
+if [ "$LAN_MODE" = 'true' ] && [ -n "$LAN_PRIVATE_IP" ]; then
   # Check if the LAN_PRIVATE_IP is reachable
   if ! ping -c 1 -W 1 "$LAN_PRIVATE_IP" > /dev/null; then
     echo "ERROR: LAN_PRIVATE_IP $LAN_PRIVATE_IP is not reachable"
@@ -26,15 +33,16 @@ else
     echo "        - MacOS: ./configure_lan_private_ip_macos.sh"
     echo "        - Windows: .\configure_lan_private_ip_windows.bat"
     echo ""
+    echo "    If you don't want to access OpenVidu through your LAN,"
+    echo "    you can run without LAN_MODE enabled, simply set"
+    echo "    the following variables in the .env file:"
+    echo "        USE_HTTPS=false"
+    echo "        LAN_MODE=false"
+    echo ""
     exit 1
   fi
 fi
 
-if [ "$LAN_MODE" = 'true' ] && [ "$USE_HTTPS" = 'false' ]; then
-  echo 'LAN_MODE cannot be "true" if USE_HTTPS is "false"'
-  exit 1
-fi
-
 # Prepare volumes
 mkdir -p /minio/data
 mkdir -p /mongo/data

pro/.env

@@ -30,9 +30,5 @@ MINIO_SECRET_KEY=minioadmin
 MONGO_ADMIN_USERNAME=mongoadmin
 MONGO_ADMIN_PASSWORD=mongoadmin
 
-# Default App (OpenVidu Call)
-CALL_PRIVATE_ACCESS=false
-CALL_USER=
-CALL_SECRET=
-CALL_ADMIN_USER=admin
-CALL_ADMIN_SECRET=admin
+# OpenVidu Meet base path
+MEET_BASE_PATH=/meet

pro/agent-speech-processing.yaml

@@ -0,0 +1,437 @@
+# Docker image of the agent.
+docker_image: docker.io/openvidu/agent-speech-processing-vosk:3.6.0
+
+# Whether to run the agent or not.
+enabled: false
+
+# Maximum CPU load threshold for the agent to accept new jobs. Value between 0 and 1.
+load_threshold: 1.0
+
+# Log level for the agent [DEBUG, INFO, WARN, ERROR, CRITICAL]
+log_level: INFO
+
+live_captions:
+  # How this agent will connect to Rooms [manual, automatic]
+  # - manual: the agent will connect to new Rooms only when your application dictates it by using the Agent Dispatch API.
+  # - automatic: the agent will automatically connect to new Rooms.
+  processing: manual
+
+  # Which speech-to-text AI provider to use [aws, azure, google, openai, azure_openai, groq, deepgram, assemblyai, fal, clova, speechmatics, gladia, sarvam, mistralai, cartesia, soniox, nvidia, elevenlabs, simplismart, vosk, sherpa]
+  # The custom configuration for the selected provider must be set below
+  provider: vosk
+
+  aws:
+    # Credentials for AWS Transcribe. See https://docs.aws.amazon.com/transcribe/latest/dg/what-is.html
+    aws_access_key_id:
+    aws_secret_access_key:
+    aws_default_region:
+    # See https://docs.aws.amazon.com/transcribe/latest/dg/supported-languages.html
+    language:
+    # The name of the custom vocabulary you want to use.
+    # See https://docs.aws.amazon.com/transcribe/latest/dg/custom-vocabulary.html
+    vocabulary_name:
+    # The name of the custom language model you want to use.
+    # See https://docs.aws.amazon.com/transcribe/latest/dg/custom-language-models-using.html
+    language_model_name:
+    # Whether or not to enable partial result stabilization. Partial result stabilization can reduce latency in your output, but may impact accuracy.
+    # See https://docs.aws.amazon.com/transcribe/latest/dg/streaming-partial-results.html#streaming-partial-result-stabilization
+    enable_partial_results_stabilization:
+    # Specify the level of stability to use when you enable partial results stabilization (enable_partial_results_stabilization: true). Valid values: high | medium | low
+    # See https://docs.aws.amazon.com/transcribe/latest/dg/streaming-partial-results.html#streaming-partial-result-stabilization
+    partial_results_stability:
+    # The name of the custom vocabulary filter you want to use to mask or remove words.
+    # See https://docs.aws.amazon.com/transcribe/latest/dg/vocabulary-filtering.html
+    vocab_filter_name:
+    # The method used to filter the vocabulary. Valid values: mask | remove | tag
+    # See https://docs.aws.amazon.com/transcribe/latest/dg/vocabulary-filtering.html
+    vocab_filter_method:
+
+  azure:
+    # Credentials for Azure Speech Service.
+    # One of these combinations must be set:
+    #  - speech_host
+    #  - speech_key + speech_region
+    #  - speech_auth_token + speech_region
+    # See https://learn.microsoft.com/en-us/azure/ai-services/speech-service/get-started-speech-to-text?tabs=macos%2Cterminal&pivots=programming-language-python#prerequisites
+    speech_host:
+    speech_key:
+    speech_auth_token:
+    speech_region:
+    # Azure handles multiple languages and can auto-detect the language used. It requires the candidate set to be set. E.g. ["en-US", "es-ES"]
+    # See https://learn.microsoft.com/en-us/azure/ai-services/speech-service/language-support?tabs=stt#supported-languages
+    language:
+    # Removes profanity (swearing), or replaces letters of profane words with stars. Valid values: Masked | Removed | Raw
+    # See https://learn.microsoft.com/en-us/azure/ai-services/translator/profanity-filtering
+    profanity:
+    # List of words or phrases to boost recognition accuracy. Azure will give higher priority to these phrases during recognition.
+    phrase_list:
+    # Controls punctuation behavior. If True, enables explicit punctuation mode where punctuation marks are added explicitly. If False (default), uses Azure's default punctuation behavior.
+    explicit_punctuation:
+
+  azure_openai:
+    # Credentials for Azure OpenAI APIs. See https://learn.microsoft.com/en-us/azure/api-management/api-management-authenticate-authorize-azure-openai
+    # Azure OpenAI API key
+    azure_api_key:
+    # Azure Active Directory token
+    azure_ad_token:
+    # Azure OpenAI endpoint in the following format: https://{your-resource-name}.openai.azure.com. Mandatory value.
+    azure_endpoint:
+    # Name of your model deployment. If given with `azure_endpoint`, sets the base client URL to include `/deployments/{azure_deployment}`.
+    azure_deployment:
+    # OpenAI REST API version used for the request. Mandatory value.
+    api_version:
+    # OpenAI organization ID.
+    organization:
+    # OpenAI project ID.
+    project:
+    # The language code to use for transcription (e.g., "en" for English).
+    language:
+    # Whether to automatically detect the language.
+    detect_language:
+    # ID of the model to use for speech-to-text.
+    model:
+    # Initial prompt to guide the transcription.
+    prompt:
+
+  google:
+    # Credentials for Google Cloud. This is the content of a Google Cloud credential JSON file.
+    # Below is a dummy example for a credential type of "Service Account" (https://cloud.google.com/iam/docs/service-account-creds#key-types)
+    credentials_info: |
+      {
+        "type": "service_account",
+        "project_id": "my-project",
+        "private_key_id": "xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx",
+        "private_key": "-----BEGIN PRIVATE KEY-----\nxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx\n-----END PRIVATE KEY-----\n",
+        "client_email": "my-email@my-project.iam.gserviceaccount.com",
+        "client_id": "xxxxxxxxxxxxxxxxxxxxx",
+        "auth_uri": "https://accounts.google.com/o/oauth2/auth",
+        "token_uri": "https://oauth2.googleapis.com/token",
+        "auth_provider_x509_cert_url": "https://www.googleapis.com/oauth2/v1/certs",
+        "client_x509_cert_url": "https://www.googleapis.com/robot/v1/metadata/x509/my-email%40my-project.iam.gserviceaccount.com",
+        "universe_domain": "googleapis.com"
+      }
+    # Which model to use for recognition. If not set, uses the default model for the selected language.
+    # See https://cloud.google.com/speech-to-text/docs/transcription-model
+    model:
+    # The location to use for recognition. Default is "us-central1". Latency will be best if the location is close to your users.
+    # Check supported languages and locations at https://cloud.google.com/speech-to-text/v2/docs/speech-to-text-supported-languages
+    location:
+    # List of language codes to recognize. Default is ["en-US"].
+    # See https://cloud.google.com/speech-to-text/v2/docs/speech-to-text-supported-languages
+    languages:
+    # Whether to detect the language of the audio. Default is true.
+    detect_language:
+    # If 'true', adds punctuation to recognition result hypotheses. This feature is only available in select languages. Setting this
+    # for requests in other languages has no effect at all. The default 'false' value does not add punctuation to result hypotheses.
+    # See https://cloud.google.com/speech-to-text/docs/automatic-punctuation
+    punctuate:
+    # The spoken punctuation behavior for the call. If not set, uses default behavior based on model of choice.
+    # e.g. command_and_search will enable spoken punctuation by default. If 'true', replaces spoken punctuation
+    # with the corresponding symbols in the request. For example, "how are you question mark" becomes "how are you?".
+    # See https://cloud.google.com/speech-to-text/docs/spoken-punctuation for support. If 'false', spoken punctuation is not replaced.
+    spoken_punctuation:
+    # Whether to return interim (non-final) transcription results. Defaults to true.
+    interim_results:
+
+  openai:
+    # API key for OpenAI. See https://platform.openai.com/api-keys
+    api_key:
+    # The OpenAI model to use for transcription. See https://platform.openai.com/docs/guides/speech-to-text
+    model:
+    # The language of the input audio. Supplying the input language in ISO-639-1 format
+    # (https://en.wikipedia.org/wiki/List_of_ISO_639-1_codes) will improve accuracy and latency.
+    language:
+    # Whether to automatically detect the language.
+    detect_language:
+    # Optional text prompt to guide the transcription. Only supported for whisper-1.
+    prompt:
+
+  groq:
+    # API key for Groq. See https://console.groq.com/keys
+    api_key:
+    # See https://console.groq.com/docs/speech-to-text
+    model:
+    # The language of the input audio. Supplying the input language in ISO-639-1 format
+    # (https://en.wikipedia.org/wiki/List_of_ISO_639-1_codes) will improve accuracy and latency.
+    language:
+    # Whether to automatically detect the language.
+    detect_language:
+    # Prompt to guide the model's style or specify how to spell unfamiliar words. 224 tokens max.
+    prompt:
+    # Base URL for the Groq API. By default "https://api.groq.com/openai/v1"
+    base_url:
+
+  deepgram:
+    # See https://console.deepgram.com/
+    api_key:
+    # See https://developers.deepgram.com/reference/speech-to-text-api/listen-streaming#request.query.model
+    model:
+    # See https://developers.deepgram.com/reference/speech-to-text-api/listen-streaming#request.query.language
+    language:
+    # Whether to enable automatic language detection. See https://developers.deepgram.com/docs/language-detection
+    detect_language: false
+    # Whether to return interim (non-final) transcription results. See https://developers.deepgram.com/docs/interim-results
+    interim_results: true
+    # Whether to apply smart formatting to numbers, dates, etc. See https://developers.deepgram.com/docs/smart-format
+    smart_format: false
+    # When smart_format is used, ensures it does not wait for sequence to be complete before returning results. See https://developers.deepgram.com/docs/smart-format#using-no-delay
+    no_delay: true
+    # Whether to add punctuations to the transcription. Turn detector will work better with punctuations. See https://developers.deepgram.com/docs/punctuation
+    punctuate: true
+    # Whether to include filler words (um, uh, etc.) in transcription. See https://developers.deepgram.com/docs/filler-words
+    filler_words: true
+    # Whether to filter profanity from the transcription. See https://developers.deepgram.com/docs/profanity-filter
+    profanity_filter: false
+    # Whether to transcribe numbers as numerals. See https://developers.deepgram.com/docs/numerals
+    numerals: false
+    # List of tuples containing keywords and their boost values for improved recognition. Each tuple should be (keyword: str, boost: float). keywords does not work with Nova-3 models. Use keyterms instead.
+    # keywords:
+    #   - [OpenVidu, 1.5]
+    #   - [WebRTC, 1]
+    # List of key terms to improve recognition accuracy. keyterms is supported by Nova-3 models.
+    # Commented below is an example
+    keyterms:
+      # - "OpenVidu"
+      # - "WebRTC"
+
+  assemblyai:
+    # API key for AssemblyAI. See https://www.assemblyai.com/dashboard/api-keys
+    api_key:
+    # The confidence threshold (0.0 to 1.0) to use when determining if the end of a turn has been reached.
+    end_of_turn_confidence_threshold:
+    # The minimum amount of silence in milliseconds required to detect end of turn when confident.
+    min_end_of_turn_silence_when_confident:
+    # The maximum amount of silence in milliseconds allowed in a turn before end of turn is triggered.
+    max_turn_silence:
+    # Whether to return formatted final transcripts (proper punctuation, letter casing...). If enabled, formatted final transcripts are emitted shortly following an end-of-turn detection.
+    format_turns: true
+    # List of keyterms to improve recognition accuracy for specific words and phrases.
+    keyterms_prompt:
+      # - "OpenVidu"
+      # - "WebRTC"
+
+  fal:
+    # API key for fal. See https://fal.ai/dashboard/keys
+    api_key:
+    # See https://fal.ai/models/fal-ai/wizper/api#schema
+    language:
+
+  clova:
+    # Secret key issued when registering the app
+    api_key:
+    # API Gateway's unique invoke URL created in CLOVA Speech Domain.
+    # See https://guide.ncloud-docs.com/docs/en/clovaspeech-domain#create-domain
+    invoke_url:
+    # See https://api.ncloud-docs.com/docs/en/ai-application-service-clovaspeech-longsentence
+    language:
+    # Value between 0 and 1 indicating the threshold for the confidence score of the transcribed text. Default is 0.5.
+    # If the confidence score is lower than the threshold, the transcription event is not sent to the client.
+    # For a definition of the confidence score see https://api.ncloud-docs.com/docs/en/ai-application-service-clovaspeech-grpc
+    threshold:
+
+  speechmatics:
+    # API key for Speechmatics. See https://portal.speechmatics.com/manage-access/
+    api_key:
+    # ISO 639-1 language code. All languages are global and can understand different dialects/accents. To see the list of all supported languages, see https://docs.speechmatics.com/speech-to-text/languages#transcription-languages
+    language:
+    # Operating point to use for the transcription per required accuracy & complexity. To learn more, see https://docs.speechmatics.com/speech-to-text/languages#operating-points
+    operating_point:
+    # Partial transcripts allow you to receive preliminary transcriptions and update as more context is available until the higher-accuracy final transcript is returned. Partials are returned faster but without any post-processing such as formatting. See https://docs.speechmatics.com/speech-to-text/realtime/output#partial-transcripts
+    enable_partials:
+    # Enable speaker diarization. When enabled, the STT engine will determine and attribute words to unique speakers. The speaker_sensitivity parameter can be used to adjust the sensitivity of diarization
+    enable_diarization:
+    # RFC-5646 language code to make spelling rules more consistent in the transcription output. See https://docs.speechmatics.com/features/word-tagging#output-locale
+    output_locale:
+    # The delay in seconds between the end of a spoken word and returning the final transcript results. See https://docs.speechmatics.com/features/realtime-latency#configuration-example
+    max_delay:
+    # See https://docs.speechmatics.com/features/realtime-latency#configuration-example
+    max_delay_mode:
+    # Configuration for speaker diarization. See https://docs.speechmatics.com/features/diarization
+    speaker_diarization_config:
+      # See https://docs.speechmatics.com/features/diarization#max-speakers
+      max_speakers:
+      # See https://docs.speechmatics.com/features/diarization#speaker-sensitivity
+      speaker_sensitivity:
+      # See https://docs.speechmatics.com/features/diarization#prefer-current-speaker
+      prefer_current_speaker:
+    # Permitted punctuation marks for advanced punctuation. See https://docs.speechmatics.com/features/punctuation-settings
+    # Commented is an example of punctuation settings
+    punctuation_overrides:
+      # permitted_marks: [ ".", "," ]
+      # sensitivity: 0.4
+    # See https://docs.speechmatics.com/features/custom-dictionary
+    # Commented below is an example of a custom dictionary
+    additional_vocab:
+      # - content: financial crisis
+      # - content: gnocchi
+      #   sounds_like:
+      #     - nyohki
+      #     - nokey
+      #     - nochi
+      # - content: CEO
+      #   sounds_like:
+      #     - C.E.O.
+
+  gladia:
+    # API key for Gladia. See https://app.gladia.io/account
+    api_key:
+    # Whether to return interim (non-final) transcription results. Defaults to True
+    interim_results:
+    # List of language codes to use for recognition. Defaults to None (auto-detect). See https://docs.gladia.io/chapters/limits-and-specifications/languages
+    languages:
+    # Whether to allow switching between languages during recognition. Defaults to True
+    code_switching:
+    # https://docs.gladia.io/api-reference/v2/live/init#body-pre-processing-audio-enhancer
+    pre_processing_audio_enhancer:
+    # https://docs.gladia.io/api-reference/v2/live/init#body-pre-processing-speech-threshold
+    pre_processing_speech_threshold:
+
+  sarvam:
+    # API key for Sarvam. See https://dashboard.sarvam.ai/key-management
+    api_key:
+    # BCP-47 language code for supported Indian languages. See https://docs.sarvam.ai/api-reference-docs/speech-to-text/transcribe#request.body.language_code.language_code
+    language:
+    # The Sarvam STT model to use. See https://docs.sarvam.ai/api-reference-docs/speech-to-text/transcribe#request.body.model.model
+    model:
+
+  mistralai:
+    # API key for Mistral AI. See https://console.mistral.ai/api-keys
+    api_key:
+    # Name of the Voxtral STT model to use. Default to voxtral-mini-latest. See https://docs.mistral.ai/capabilities/audio/
+    model:
+    # The language code to use for transcription (e.g., "en" for English)
+    language:
+
+  cartesia:
+    # API key for Cartesia. See https://play.cartesia.ai/keys
+    api_key:
+    # The Cartesia STT model to use
+    model:
+    # The language code to use for transcription (e.g., "en" for English)
+    language:
+
+  soniox:
+    # API key for Soniox. See https://console.soniox.com/
+    api_key:
+    # Set language hints when possible to significantly improve accuracy. See: https://soniox.com/docs/stt/concepts/language-hints
+    language_hints:
+      # - "en"
+      # - "es"
+    # Set context to improve recognition of difficult and rare words. Context is a string and can include words, phrases, sentences, or summaries (limit: 10K chars). See https://soniox.com/docs/stt/concepts/context
+    context:
+
+  nvidia:
+    # API key for NVIDIA. See https://build.nvidia.com/explore/speech?integrate_nim=true&hosted_api=true&modal=integrate-nim
+    # Required when using NVIDIA's cloud services. To use a self-hosted NVIDIA Riva server setup "server" and "use_ssl" instead.
+    api_key:
+    # The NVIDIA Riva ASR model to use. Default is "parakeet-1.1b-en-US-asr-streaming-silero-vad-sortformer"
+    # See available models: https://build.nvidia.com/search/models?filters=usecase%3Ausecase_speech_to_text
+    model:
+    # The NVIDIA function ID for the model. Default is "1598d209-5e27-4d3c-8079-4751568b1081"
+    function_id:
+    # Whether to add punctuation to transcription results. Default is true.
+    punctuate:
+    # The language code for transcription. Default is "en-US"
+    language_code:
+    # Audio sample rate in Hz. Default is 16000.
+    sample_rate:
+    # The NVIDIA Riva server address. Default is "grpc.nvcf.nvidia.com:443"
+    # For self-hosted NIM, use your server address (e.g., "localhost:50051")
+    server:
+    # Whether to use SSL for the connection. Default is true.
+    # Set to false for locally hosted Riva NIM services without SSL.
+    use_ssl:
+
+  spitch:
+    # API key for Spitch. See https://docs.spitch.app/keys
+    api_key:
+    # Language short code for the generated speech. For supported values, see https://docs.spitch.app/concepts/languages
+    language:
+
+  elevenlabs:
+    # API key for ElevenLabs. See https://elevenlabs.io/app/settings/api-keys
+    api_key:
+    # The ElevenLabs STT model to use. Valid values are ["scribe_v1", "scribe_v2", "scribe_v2_realtime"]. See https://elevenlabs.io/docs/overview/models#models-overview
+    model_id:
+    # An ISO-639-1 or ISO-639-3 language_code corresponding to the language of the audio file. Can sometimes improve transcription performance if known beforehand. Defaults to null, in this case the language is predicted automatically
+    language_code:
+    # Custom base URL for the API. Optional.
+    base_url:
+    # Audio sample rate in Hz. Default is 16000.
+    sample_rate:
+    # Whether to tag audio events like (laughter), (footsteps), etc. in the transcription. Only supported for Scribe v1 model. Default is True
+    tag_audio_events:
+    # Whether to include word-level timestamps in the transcription. Default is false.
+    include_timestamps:
+
+  simplismart:
+    # API key for SimpliSmart. See https://docs.simplismart.ai/model-suite/settings/api-keys
+    api_key:
+    # Model identifier for the backend STT model. One of ["openai/whisper-large-v2", "openai/whisper-large-v3", "openai/whisper-large-v3-turbo"]
+    # Default is "openai/whisper-large-v3-turbo"
+    model:
+    # Language code for transcription (default: "en"). See https://docs.simplismart.ai/get-started/playground/transcription-models#supported-languages-with-their-codes
+    language:
+    # Operation to perform. "transcribe" converts speech to text in the original language, "translate" translates into English. Default is "transcribe".
+    task:
+    # If true, disables timestamp generation in transcripts. Default is true
+    without_timestamps:
+    # Minimum duration (ms) for a valid speech segment. Default is 0
+    min_speech_duration_ms:
+    # Decoding temperature (affects randomness). Default is 0.0
+    temperature:
+    # Whether to permit multilingual recognition. Default is false
+    multilingual:
+
+  vosk:
+    # Vosk language model. This provider requires docker_image "docker.io/openvidu/agent-speech-processing-vosk"
+    # Below is the list of pre-installed models in the container (available at https://alphacephei.com/vosk/models):
+    # - vosk-model-en-us-0.22-lgraph (English US)
+    # - vosk-model-small-cn-0.22 (Chinese)
+    # - vosk-model-small-de-0.15 (German)
+    # - vosk-model-small-en-in-0.4 (English India)
+    # - vosk-model-small-es-0.42 (Spanish)
+    # - vosk-model-small-fr-0.22 (French)
+    # - vosk-model-small-hi-0.22 (Hindi)
+    # - vosk-model-small-it-0.22 (Italian)
+    # - vosk-model-small-ja-0.22 (Japanese)
+    # - vosk-model-small-nl-0.22 (Dutch)
+    # - vosk-model-small-pt-0.3 (Portuguese)
+    # - vosk-model-small-ru-0.22 (Russian)
+    model: vosk-model-en-us-0.22-lgraph
+    # Language code for reference. It has no effect other than observability purposes.
+    # If a pre-installed "model" is declared, this will be set automatically if empty.
+    language:
+    # Audio sample rate in Hz. Default is 16000.
+    sample_rate:
+    # Whether to return interim/partial results during recognition. Default is true.
+    partial_results:
+    # Whether to override Vosk's built-in Voice Activity Detection (VAD) with Silero's VAD. Default is false.
+    use_silero_vad: false
+
+  sherpa:
+    # sherpa streaming model. This provider requires docker_image "docker.io/openvidu/agent-speech-processing-sherpa"
+    # Below is the list of pre-installed models in the container (available at https://github.com/k2-fsa/sherpa-onnx/releases/tag/asr-models):
+    # - sherpa-onnx-streaming-zipformer-en-kroko-2025-08-06 (English)
+    # - sherpa-onnx-streaming-zipformer-es-kroko-2025-08-06 (Spanish)
+    # - sherpa-onnx-streaming-zipformer-de-kroko-2025-08-06 (German)
+    # - sherpa-onnx-streaming-zipformer-fr-kroko-2025-08-06 (French)
+    # - sherpa-onnx-streaming-zipformer-ar_en_id_ja_ru_th_vi_zh-2025-02-10 (Multilingual: Arabic, English, Indonesian, Japanese, Russian, Thai, Vietnamese, Chinese)
+    model: sherpa-onnx-streaming-zipformer-en-kroko-2025-08-06
+    # Language code for reference. Auto-detected from model name if not set.
+    language:
+    # Runtime provider for sherpa-onnx. Supported values: "cpu" or "cuda". Default is "cpu".
+    # Learn about GPU acceleration at https://openvidu.io/docs/ai/live-captions/#gpu-acceleration-for-sherpa-provider
+    provider:
+    # Audio sample rate in Hz. Default is 16000.
+    sample_rate:
+    # Whether to return interim/partial results during recognition. Default is true.
+    partial_results:
+    # Number of threads for ONNX Runtime. Default is 2.
+    num_threads:
+    # Recognizer type ("transducer", "paraformer", "zipformer_ctc", "nemo_ctc", "t_one_ctc"). Auto-detected from model name if not set.
+    recognizer_type:
+    # Decoding method ("greedy_search", "modified_beam_search"). Default is "greedy_search".
+    decoding_method:
+    # Whether to override sherpa's built-in Voice Activity Detection (VAD) with Silero's VAD. Default is false.
+    use_silero_vad: false

pro/configure_lan_private_ip_macos.sh

@@ -14,14 +14,4 @@ if [ -z "$LAN_PRIVATE_IP" ]; then
 fi
 
 # Replace the LAN_PRIVATE_IP in the .env file
-sed -i'' -e "s/LAN_PRIVATE_IP=.*/LAN_PRIVATE_IP=$LAN_PRIVATE_IP/g" .env
-
-# If sillicon mac, enable EXPERIMENTAL_DOCKER_DESKTOP_FORCE_QEMU flag
-if [ "$(uname -m)" = "arm64" ]; then
-    if ! grep -q "EXPERIMENTAL_DOCKER_DESKTOP_FORCE_QEMU" .env; then
-        echo "# Enable this flag to run Docker Desktop on Apple Silicon Macs" >> .env
-        echo "EXPERIMENTAL_DOCKER_DESKTOP_FORCE_QEMU=1" >> .env
-    else
-        sed -i'' -e "s/EXPERIMENTAL_DOCKER_DESKTOP_FORCE_QEMU=.*/EXPERIMENTAL_DOCKER_DESKTOP_FORCE_QEMU=1/g" .env
-    fi
-fi
+sed -i'' -e "s/LAN_PRIVATE_IP=.*/LAN_PRIVATE_IP=$LAN_PRIVATE_IP/g" .env

pro/docker-compose.yaml

@@ -1,7 +1,6 @@
 services:
   caddy-proxy:
-    image: docker.io/openvidu/openvidu-caddy-local:3.1.0
-    platform: linux/amd64
+    image: docker.io/openvidu/openvidu-caddy-local:3.6.0
     container_name: caddy-proxy
     restart: unless-stopped
     extra_hosts:
@@ -18,62 +17,67 @@ services:
       - MINIO_ACCESS_KEY=${MINIO_ACCESS_KEY:-}
       - MINIO_SECRET_KEY=${MINIO_SECRET_KEY:-}
       - V2COMPAT_OPENVIDU_SECRET=${LIVEKIT_API_SECRET:-}
+      - MEET_BASE_PATH=${MEET_BASE_PATH:-/meet}
+    env_file:
+      - ./meet.env
     volumes:
       - ./custom-layout:/var/www/custom-layout
+      - /etc/localtime:/etc/localtime:ro
     ports:
       - 5443:5443
       - 6443:6443
       - 7443:7443
       - 7880:7880
+      - 9443:9443
+      - 9080:9080
     depends_on:
       setup:
         condition: service_completed_successfully
 
   redis:
-    image: docker.io/redis:7.4.2-alpine
-    platform: linux/amd64
+    image: docker.io/redis:8.6.1-alpine
     container_name: redis
     restart: unless-stopped
     ports:
       - 6379:6379
     volumes:
       - redis:/data
+      - /etc/localtime:/etc/localtime:ro
     command: >
-      redis-server
-      --bind 0.0.0.0
-      --requirepass ${REDIS_PASSWORD:-}
+      redis-server --bind 0.0.0.0 --requirepass ${REDIS_PASSWORD:-}
     depends_on:
       setup:
         condition: service_completed_successfully
 
   minio:
-    image: docker.io/bitnami/minio:2025.2.7-debian-12-r0
-    platform: linux/amd64
+    image: docker.io/openvidu/minio:2025.10.15-debian-12-r9
     restart: unless-stopped
     ports:
       - 9000:9000
     environment:
       - MINIO_ROOT_USER=${MINIO_ACCESS_KEY:-}
       - MINIO_ROOT_PASSWORD=${MINIO_SECRET_KEY:-}
-      - MINIO_DEFAULT_BUCKETS=openvidu
+      - MINIO_DEFAULT_BUCKETS=openvidu-appdata
       - MINIO_CONSOLE_SUBPATH=/minio-console
+      - MINIO_BROWSER=on
       - MINIO_BROWSER_REDIRECT_URL=http://localhost:7880/minio-console
     volumes:
       - minio-data:/bitnami/minio/data
       - minio-certs:/certs
+      - /etc/localtime:/etc/localtime:ro
     depends_on:
       setup:
         condition: service_completed_successfully
 
   mongo:
-    image: docker.io/bitnami/mongodb:8.0.4
-    platform: linux/amd64
+    image: docker.io/openvidu/mongodb:8.0.19-r1
     container_name: mongo
     restart: unless-stopped
     ports:
       - 27017:27017
     volumes:
       - mongo-data:/bitnami/mongodb
+      - /etc/localtime:/etc/localtime:ro
     environment:
       - MONGODB_ROOT_USER=${MONGO_ADMIN_USERNAME:-}
       - MONGODB_ROOT_PASSWORD=${MONGO_ADMIN_PASSWORD:-}
@@ -81,14 +85,12 @@ services:
       - MONGODB_REPLICA_SET_MODE=primary
       - MONGODB_REPLICA_SET_NAME=rs0
       - MONGODB_REPLICA_SET_KEY=devreplicasetkey
-      - EXPERIMENTAL_DOCKER_DESKTOP_FORCE_QEMU=${EXPERIMENTAL_DOCKER_DESKTOP_FORCE_QEMU:-0}
     depends_on:
       setup:
         condition: service_completed_successfully
 
   dashboard:
-    image: docker.io/openvidu/openvidu-dashboard:3.1.0
-    platform: linux/amd64
+    image: docker.io/openvidu/openvidu-dashboard:3.6.0
     container_name: dashboard
     restart: unless-stopped
     environment:
@@ -96,18 +98,20 @@ services:
       - ADMIN_USERNAME=${DASHBOARD_ADMIN_USERNAME:-}
       - ADMIN_PASSWORD=${DASHBOARD_ADMIN_PASSWORD:-}
       - DATABASE_URL=mongodb://${MONGO_ADMIN_USERNAME}:${MONGO_ADMIN_PASSWORD}@mongo:27017/?replicaSet=rs0&readPreference=primaryPreferred
+    volumes:
+      - /etc/localtime:/etc/localtime:ro
     depends_on:
       setup:
         condition: service_completed_successfully
 
   openvidu:
-    image: docker.io/openvidu/openvidu-server-pro:3.1.0
-    platform: linux/amd64
+    image: docker.io/openvidu/openvidu-server-pro:3.6.0
     restart: unless-stopped
     container_name: openvidu
     extra_hosts:
       - host.docker.internal:host-gateway
     environment:
+      - LAN_MODE=${LAN_MODE:-false}
       - LAN_PRIVATE_IP=${LAN_PRIVATE_IP:-}
       - OPENVIDU_DEPLOYMENT_TYPE=local
       - OPENVIDU_ENVIRONMENT=on_premise
@@ -120,15 +124,17 @@ services:
     volumes:
       - ./livekit.yaml:/etc/livekit.yaml
       - ./scripts/entrypoint_openvidu.sh:/scripts/entrypoint.sh
+      - /etc/localtime:/etc/localtime:ro
     depends_on:
       setup:
         condition: service_completed_successfully
 
   ingress:
-    image: docker.io/openvidu/ingress:3.1.0
-    platform: linux/amd64
+    image: docker.io/openvidu/ingress:3.6.0
     container_name: ingress
     restart: unless-stopped
+    extra_hosts:
+      - host.docker.internal:host-gateway
     ports:
       - 1935:1935
       - 8085:8085
@@ -137,13 +143,13 @@ services:
       - INGRESS_CONFIG_FILE=/etc/ingress.yaml
     volumes:
       - ./ingress.yaml:/etc/ingress.yaml
+      - /etc/localtime:/etc/localtime:ro
     depends_on:
       setup:
         condition: service_completed_successfully
 
   egress:
-    image: docker.io/livekit/egress:v1.9.0
-    platform: linux/amd64
+    image: docker.io/openvidu/egress:3.6.0
     restart: unless-stopped
     container_name: egress
     extra_hosts:
@@ -153,49 +159,42 @@ services:
     volumes:
       - ./egress.yaml:/etc/egress.yaml
       - egress-data:/home/egress
+      - /etc/localtime:/etc/localtime:ro
     depends_on:
       setup:
         condition: service_completed_successfully
 
-  default-app:
-    image: docker.io/openvidu/openvidu-call:3.1.0-demo
-    platform: linux/amd64
-    container_name: openvidu-call
+  openvidu-meet:
+    image: docker.io/openvidu/openvidu-meet:3.6.0
+    container_name: openvidu-meet
     restart: on-failure
+    extra_hosts:
+      - host.docker.internal:host-gateway
     environment:
       - USE_HTTPS=${USE_HTTPS:-false}
       - LAN_MODE=${LAN_MODE:-false}
       - LAN_DOMAIN=${LAN_DOMAIN:-}
       - LAN_PRIVATE_IP=${LAN_PRIVATE_IP:-}
-      - SERVER_PORT=6080
-      - CALL_NAME_ID=OpenViduCall-LOCAL
-      - LIVEKIT_URL_PRIVATE=ws://openvidu:7880/
       - LIVEKIT_API_KEY=${LIVEKIT_API_KEY}
       - LIVEKIT_API_SECRET=${LIVEKIT_API_SECRET}
-      - CALL_PRIVATE_ACCESS=${CALL_PRIVATE_ACCESS:-false}
-      - CALL_USER=${CALL_USER:-}
-      - CALL_SECRET=${CALL_SECRET:-}
-      - CALL_RECORDING=${CALL_RECORDING:-}
-      - CALL_ADMIN_USER=${CALL_ADMIN_USER:-admin}
-      - CALL_ADMIN_SECRET=${CALL_ADMIN_SECRET:-admin}
-      - CALL_LOG_LEVEL=${CALL_LOG_LEVEL:-info}
-      - CALL_S3_BUCKET=${CALL_S3_BUCKET:-openvidu}
-      - CALL_S3_SERVICE_ENDPOINT=${CALL_S3_SERVICE_ENDPOINT:-http://minio:9000}
-      - CALL_S3_ACCESS_KEY=${MINIO_ACCESS_KEY}
-      - CALL_S3_SECRET_KEY=${MINIO_SECRET_KEY}
-      - CALL_AWS_REGION=${CALL_AWS_REGION:-us-east-1}
-      - CALL_S3_WITH_PATH_STYLE_ACCESS=${CALL_S3_WITH_PATH_STYLE_ACCESS:-true}
+      - MEET_S3_ACCESS_KEY=${MINIO_ACCESS_KEY}
+      - MEET_S3_SECRET_KEY=${MINIO_SECRET_KEY}
+      - MEET_REDIS_PASSWORD=${REDIS_PASSWORD:-}
+      - MEET_MONGO_URI=mongodb://${MONGO_ADMIN_USERNAME}:${MONGO_ADMIN_PASSWORD}@mongo:27017/?replicaSet=rs0&readPreference=primaryPreferred
+      - MEET_BASE_PATH=${MEET_BASE_PATH:-/meet}
+      - MEET_CONFIG_DIR=/config/meet.env
     volumes:
-      - ./scripts/entrypoint_default_app.sh:/scripts/entrypoint.sh
+      - ./meet.env:/config/meet.env
+      - ./scripts/entrypoint_openvidu_meet.sh:/scripts/entrypoint.sh
       - ./scripts/utils.sh:/scripts/utils.sh
+      - /etc/localtime:/etc/localtime:ro
     entrypoint: /bin/sh /scripts/entrypoint.sh
     depends_on:
       setup:
         condition: service_completed_successfully
 
   openvidu-v2compatibility:
-    image: docker.io/openvidu/openvidu-v2compatibility:3.1.0
-    platform: linux/amd64
+    image: docker.io/openvidu/openvidu-v2compatibility:3.6.0
     restart: unless-stopped
     container_name: openvidu-v2compatibility
     entrypoint: /bin/sh /scripts/entrypoint.sh
@@ -215,7 +214,7 @@ services:
       - V2COMPAT_LIVEKIT_API_SECRET=${LIVEKIT_API_SECRET:-}
       - V2COMPAT_OPENVIDU_RECORDING_PATH=/opt/openvidu/recordings
       - V2COMPAT_OPENVIDU_PRO_RECORDING_STORAGE=local
-      - V2COMPAT_OPENVIDU_PRO_AWS_S3_BUCKET=openvidu
+      - V2COMPAT_OPENVIDU_PRO_AWS_S3_BUCKET=openvidu-appdata
       - V2COMPAT_OPENVIDU_PRO_AWS_S3_SERVICE_ENDPOINT=http://minio:9000
       - V2COMPAT_OPENVIDU_PRO_AWS_REGION=us-east-1
       - V2COMPAT_OPENVIDU_PRO_AWS_ACCESS_KEY=${MINIO_ACCESS_KEY:-}
@@ -231,27 +230,33 @@ services:
       - ./recordings:/opt/openvidu/recordings
       - ./scripts/entrypoint_v2comp.sh:/scripts/entrypoint.sh
       - ./scripts/utils.sh:/scripts/utils.sh
+      - /etc/localtime:/etc/localtime:ro
     depends_on:
       setup:
         condition: service_completed_successfully
 
   ready-check:
-    image: docker.io/curlimages/curl:8.12.0
-    platform: linux/amd64
+    image: docker.io/openvidu/openvidu-operator:3.6.0
     container_name: ready-check
     restart: on-failure
+    volumes:
+      - /etc/localtime:/etc/localtime:ro
     environment:
+      - MODE=local-ready-check
+      - OPENVIDU_ENVIRONMENT=local-platform
       - USE_HTTPS=${USE_HTTPS:-false}
       - LAN_DOMAIN=${LAN_DOMAIN:-}
       - LAN_MODE=${LAN_MODE:-false}
       - LAN_PRIVATE_IP=${LAN_PRIVATE_IP:-}
-      - V2COMPAT_OPENVIDU_SECRET=${LIVEKIT_API_SECRET:-}
-      - LIVEKIT_API_KEY=${LIVEKIT_API_KEY:-}
-      - LIVEKIT_API_SECRET=${LIVEKIT_API_SECRET:-}
       - DASHBOARD_ADMIN_USERNAME=${DASHBOARD_ADMIN_USERNAME:-}
       - DASHBOARD_ADMIN_PASSWORD=${DASHBOARD_ADMIN_PASSWORD:-}
       - MINIO_ACCESS_KEY=${MINIO_ACCESS_KEY:-}
       - MINIO_SECRET_KEY=${MINIO_SECRET_KEY:-}
+      - LIVEKIT_API_KEY=${LIVEKIT_API_KEY:-}
+      - LIVEKIT_API_SECRET=${LIVEKIT_API_SECRET:-}
+      - V2COMPAT_OPENVIDU_SECRET=${LIVEKIT_API_SECRET:-}
+    env_file:
+      - ./meet.env
     depends_on:
       - openvidu
       - ingress
@@ -259,14 +264,32 @@ services:
       - dashboard
       - minio
       - mongo
-    volumes:
-      - ./scripts/ready-check.sh:/scripts/ready-check.sh
-      - ./scripts/utils.sh:/scripts/utils.sh
-    command: /bin/sh /scripts/ready-check.sh
 
+  operator:
+    image: docker.io/openvidu/openvidu-operator:3.6.0
+    container_name: operator
+    restart: unless-stopped
+    volumes:
+      - /var/run/docker.sock:/var/run/docker.sock
+      - agents-config:/agents-config
+      - ./:/deployment
+      - /etc/localtime:/etc/localtime:ro
+    environment:
+      - MODE=agent-manager-local
+      - DEPLOYMENT_FILES_DIR=/deployment
+      - AGENTS_CONFIG_DIR=/agents-config
+      - NETWORK_NAME=openvidu-pro
+      - AGENTS_CONFIG_VOLUME=openvidu-pro-agents-config
+      - LIVEKIT_URL=ws://openvidu:7880/
+      - LIVEKIT_API_KEY=${LIVEKIT_API_KEY:-}
+      - LIVEKIT_API_SECRET=${LIVEKIT_API_SECRET:-}
+      - REDIS_ADDRESS=redis:6379
+      - REDIS_PASSWORD=${REDIS_PASSWORD:-}
+    depends_on:
+      setup:
+        condition: service_completed_successfully
   setup:
     image: docker.io/busybox:1.37.0
-    platform: linux/amd64
     container_name: setup
     restart: "no"
     volumes:
@@ -274,6 +297,7 @@ services:
       - mongo-data:/mongo
       - egress-data:/egress
       - ./scripts/setup.sh:/scripts/setup.sh
+      - /etc/localtime:/etc/localtime:ro
     environment:
       - USE_HTTPS=${USE_HTTPS:-false}
       - LAN_MODE=${LAN_MODE:-false}
@@ -283,9 +307,21 @@ services:
     command: /bin/sh /scripts/setup.sh
 
 volumes:
+  agents-config:
+    name: openvidu-pro-agents-config
   minio-certs:
+    name: openvidu-pro-minio-certs
   mongodb-config:
+    name: openvidu-pro-mongodb-config
   redis:
+    name: openvidu-pro-redis
   minio-data:
+    name: openvidu-pro-minio-data
   mongo-data:
+    name: openvidu-pro-mongo-data
   egress-data:
+    name: openvidu-pro-egress-data
+
+networks:
+  default:
+    name: openvidu-pro

pro/egress.yaml

@@ -21,12 +21,18 @@ storage:
         # Default region for minio
         region: us-east-1
         endpoint: http://minio:9000
-        bucket: openvidu
+        bucket: openvidu-appdata
         force_path_style: true
+    #azure:
+    #    account_name: your_account_name
+    #    account_key: your_account_key
+    #    container_name: openvidu-appdata
+    # gcp:
+    #     credentials_json: |
+    #         your_credentials_json
+    #     bucket: openvidu-appdata
 
 # CPU cost for each type of Egress operation.
-# Defaults are set for unlimited operations.
-# Change these values according to your needs.
 cpu_cost:
     max_cpu_utilization: 0.80
     room_composite_cpu_cost: 0.01
@@ -36,3 +42,20 @@ cpu_cost:
     participant_cpu_cost: 0.01
     track_composite_cpu_cost: 0.01
     track_cpu_cost: 0.01
+
+openvidu:
+    # Allocation strategy for new egress requests
+    # - cpuload: the node with the lowest CPU load will be selected. Distributes the CPU load evenly across all nodes.
+    # - binpack: some node already hosting at least one egress will be selected. Fills up nodes before assigning work to new ones.
+    allocation_strategy: cpuload
+    # Whether to use system-wide CPU monitoring or egress process CPU monitoring. This affects the allocation of new egress requests.
+    # It is preferable to set this value to:
+    # - true: when the egress service is running in a shared server also hosting other CPU-intensive services.
+    # - false: when the egress service is running in a dedicated server.
+    use_global_cpu_monitoring: true
+    # Disables the automatic killing of the most expensive egress when CPU is overloaded.
+    # The default "false" value helps keeping the node stable, but may cause unexpected egress terminations under high load.
+    disable_cpu_overload_killer: false
+    # Minimum available disk space in MB required to accept new egress requests.
+    # Default: 512 MB. Set to a negative value (e.g., -1) to disable disk space checking.
+    min_disk_space_mb: 512

pro/livekit.yaml

@@ -56,7 +56,7 @@ webhook:
     urls:
         - http://host.docker.internal:4443/livekit/webhook # For OpenVidu 2 compatibility
         - http://host.docker.internal:6080/livekit/webhook
-        - http://default-app:6080/livekit/webhook
+        - http://openvidu-meet:6080/livekit/webhook
 ingress:
     rtmp_base_url: rtmp://localhost:1935/rtmp
     whip_base_url: http://localhost:8085/whip

pro/meet.env

@@ -0,0 +1,36 @@
+# OpenVidu Meet configuration
+# Static environment variables loaded via MEET_CONFIG_DIR
+
+SERVER_PORT=6080
+MEET_NAME_ID=openviduMeet-LOCAL
+MEET_LOG_LEVEL=info
+MEET_COOKIE_SECURE=false
+MEET_INITIAL_ADMIN_USER=admin
+MEET_INITIAL_ADMIN_PASSWORD=admin
+MEET_INITIAL_API_KEY=meet-api-key
+MEET_INITIAL_WEBHOOK_ENABLED=true
+MEET_INITIAL_WEBHOOK_URL=http://host.docker.internal:6080/webhook
+
+LIVEKIT_URL_PRIVATE=ws://openvidu:7880/
+
+# S3 configuration
+MEET_S3_BUCKET=openvidu-appdata
+MEET_S3_SUBBUCKET=openvidu-meet
+MEET_S3_SERVICE_ENDPOINT=http://minio:9000
+MEET_AWS_REGION=us-east-1
+MEET_S3_WITH_PATH_STYLE_ACCESS=true
+
+# Storage backend type
+MEET_BLOB_STORAGE_MODE=s3
+
+# Redis configuration
+MEET_REDIS_HOST=redis
+MEET_REDIS_PORT=6379
+MEET_REDIS_DB=0
+
+# MongoDB configuration
+MEET_MONGO_ENABLED=true
+MEET_MONGO_DB_NAME=openvidu-meet
+
+# Enable live captions using OpenVidu Speech to Text agent
+MEET_CAPTIONS_ENABLED=false

pro/scripts/entrypoint_openvidu.sh

@@ -1,7 +1,8 @@
 #!/bin/sh
 set -e
 
-if [ "$LAN_PRIVATE_IP" != "none" ]; then
+if [ "$LAN_PRIVATE_IP" != "" ] && [ "$LAN_MODE" = 'true' ]; then
+    echo "Using as NODE_IP: $LAN_PRIVATE_IP"
     export NODE_IP="$LAN_PRIVATE_IP"
 fi
 

pro/scripts/entrypoint_openvidu_meet.sh

@@ -2,7 +2,7 @@
 
 . /scripts/utils.sh
 
-URL=$(getDeploymentUrl)
+URL=$(getDeploymentUrl ws)
 export LIVEKIT_URL="${URL}"
 
 /usr/local/bin/entrypoint.sh

pro/scripts/ready-check.sh

@@ -1,70 +0,0 @@
-#!/bin/sh
-
-. /scripts/utils.sh
-
-trap 'handle_sigint' SIGINT
-
-handle_sigint() {
-  echo "SIGINT signal received, exiting..."
-  exit 1
-}
-
-wait_for_service() {
-  SERVICE_NAME=$1
-  SERVICE_URL=$2
-  shift 2
-  EXTRA=$@
-  if [ -n "$EXTRA" ]; then
-    until curl $EXTRA $SERVICE_URL > /dev/null; do
-      echo "Waiting for $SERVICE_NAME to start...";
-      sleep 1;
-    done;
-  else
-    until curl --silent --head --fail $SERVICE_URL > /dev/null; do
-      echo "Waiting for $SERVICE_NAME to start...";
-      sleep 1;
-    done;
-  fi;
-}
-
-wait_for_service 'OpenVidu' 'http://openvidu:7880'
-wait_for_service 'Ingress' 'http://ingress:9091'
-wait_for_service 'Egress' 'http://egress:9091'
-wait_for_service 'Dashboard' 'http://dashboard:5000'
-wait_for_service 'Minio' 'http://minio:9000/minio/health/live'
-wait_for_service 'Minio Console' 'http://minio:9001/minio-console'
-wait_for_service 'Mongo' 'http://mongo:27017' --connect-timeout 10 --silent
-
-LAN_HTTP_URL=$(getDeploymentUrl http)
-LAN_WS_URL=$(getDeploymentUrl ws)
-
-for i in $(seq 1 10); do
-  echo 'Starting OpenVidu... Please be patient...'
-  sleep 1
-done;
-echo ''
-echo ''
-echo '========================================='
-echo '🎉 OpenVidu is ready! 🎉'
-echo '========================================='
-echo ''
-echo 'OpenVidu Server & LiveKit Server URLs:'
-echo ''
-echo '    - From this machine:'
-echo ''
-echo '        - http://localhost:7880'
-echo '        - ws://localhost:7880'
-echo ''
-echo '    - From other devices in your LAN:'
-echo ''
-echo "        - $LAN_HTTP_URL"
-echo "        - $LAN_WS_URL"
-echo ''
-echo '========================================='
-echo ''
-echo 'OpenVidu Developer UI (services and passwords):'
-echo ''
-echo '    - http://localhost:7880'
-echo "    - $LAN_HTTP_URL"
-echo ''
-echo '========================================='

pro/scripts/setup.sh

@@ -1,6 +1,11 @@
 #!/bin/sh
 
-if [ -z "$LAN_PRIVATE_IP" ]; then
+if [ "$LAN_MODE" = 'true' ] && [ "$USE_HTTPS" = 'false' ]; then
+  echo 'LAN_MODE cannot be "true" if USE_HTTPS is "false"'
+  exit 1
+fi
+
+if [ "$LAN_MODE" = 'true' ] && [ -z "$LAN_PRIVATE_IP" ]; then
   echo '------------------------'
   echo ''
   echo 'LAN_PRIVATE_IP is required in the .env file.'
@@ -14,7 +19,9 @@ if [ -z "$LAN_PRIVATE_IP" ]; then
   echo 'If it can'\''t be found, you can manually set it in the .env file'
   echo '------------------------'
   exit 1
-else
+fi
+
+if [ "$LAN_MODE" = 'true' ] && [ -n "$LAN_PRIVATE_IP" ]; then
   # Check if the LAN_PRIVATE_IP is reachable
   if ! ping -c 1 -W 1 "$LAN_PRIVATE_IP" > /dev/null; then
     echo "ERROR: LAN_PRIVATE_IP $LAN_PRIVATE_IP is not reachable"
@@ -26,15 +33,16 @@ else
     echo "        - MacOS: ./configure_lan_private_ip_macos.sh"
     echo "        - Windows: .\configure_lan_private_ip_windows.bat"
     echo ""
+    echo "    If you don't want to access OpenVidu through your LAN,"
+    echo "    you can run without LAN_MODE enabled, simply set"
+    echo "    the following variables in the .env file:"
+    echo "        USE_HTTPS=false"
+    echo "        LAN_MODE=false"
+    echo ""
     exit 1
   fi
 fi
 
-if [ "$LAN_MODE" = 'true' ] && [ "$USE_HTTPS" = 'false' ]; then
-  echo 'LAN_MODE cannot be "true" if USE_HTTPS is "false"'
-  exit 1
-fi
-
 # Prepare volumes
 mkdir -p /minio/data
 mkdir -p /mongo/data