submaster/DOCS/diagrams/high_level.mmd
Cesar Mendivil 1264ae8587 Add diagrams and documentation for the Whisper Pipeline architecture
- Created flowcharts for Infra adapters, artifacts, high-level process, models, and tests in the DOCS/diagrams directory.
- Added a mkdocs.yml configuration file for documentation site setup.
2025-10-24 16:49:49 -07:00

44 lines
1.5 KiB
Plaintext

flowchart TD
VInput["Video input"] --> ExtractAudio["Extract audio\n(ffmpeg)"]
ExtractAudio --> Transcribe["Transcription\n(faster-whisper / openai-whisper)"]
Transcribe --> Segments["Segments\n(start/end/text)"]
Segments --> Translate["Translation decision\n(usecase)"]
Translate -->|local| LocalTranslate["Local MarianMT\n(Helsinki-NLP)"]
Translate -->|remote| GeminiAPI["Gemini / LLM HTTP API"]
LocalTranslate --> SRTGen["Generate SRT\n(translated)"]
GeminiAPI --> SRTGen
SRTGen --> KokoroTTS["Kokoro TTS\n(per-segment synth)"]
KokoroTTS --> ChunkAdjust["Align / Pad / Trim\n(per segment)"]
ChunkAdjust --> Concat["Concat / Normalize WAVs"]
Concat --> AudioProc["FFmpeg Audio Proc\n(replace or mix)"]
AudioProc --> BurnSubs["Burn subtitles (ffmpeg)"]
BurnSubs --> Output["output/<basename>/\n(final artifacts)"]
%% Agrupar visualmente (se mantiene la semántica, ahora vertical)
subgraph InfraAdapters [Infra adapters]
direction TB
ExtractAudio
Transcribe
KokoroTTS
AudioProc
end
subgraph Usecase [Usecase / Orchestrator]
direction TB
Transcribe
Translate
SRTGen
Concat
end
%% servicios auxiliares (con conexiones punteadas)
Logs["Logging (logs/infra, logs/usecases, logs/cli)"]
Cache["Model cache (~/.cache/transformers)"]
Logs -.-> Transcribe
Logs -.-> KokoroTTS
Cache -.-> Transcribe
Cache -.-> LocalTranslate
classDef infra fill:#f8f9fa,stroke:#0366d6,color:#000
class InfraAdapters infra