script to lib

Refactor import statement for AudioProcessor and update cleanup method to be awaited; remove unused formatters and state management files

Refactor AudioProcessor methods for improved async handling and WebSocket integration

Refactor DiartDiarization initialization and streamline WebSocket audio processing

refacto 0

Refactor PCM conversion to a dedicated function; immediate chunk addition to the diarization queue

Warmup functionality: add timeout option (for VM not connected to internet); False option to disable warmup

fix #84

Diarization : Uses a rx observer instead of diart attach_hooks method

use confidence scores returned by whisper to immediately validate tokens

ffmpeg timout from 5 to 15s; diarization lag does not stay = 0 at the beginning

solve https://github.com/QuentinFuxa/whisper_streaming_web/issues/60#issuecomment-2692191781

update import paths

undiarized text is assigned to last speaker, with buffer information; traceback is used to format_exc errors

// execution for diarization and transcription

new buffer format

diarization now works at word - not chunk - level!

Enhance diarization logic to improve speaker attribution : corrects several bugs

DiartDiarization loaded in asynccontextmanager lifespan

Improve speaker handling; update sleep duration and manage speaker transitions more effectively

add parameter to disable transcription (only diarization), add time in output

Implement logging for WebSocket events and FFmpeg process management; remove obsolete test file

handle ffmpeg timeouts > 5s

Add get_buffer method to retrieve unvalidated buffer in string format

optimize ffmpeg buffer reading :round duration to nearest lower 0.1s

Use lifespan to load the model just one

remove logs handler

all text-related classes now share a common TimedText base class

buffer is now transcript.text

change log format

add logging config to server

Merge branch 'main' into fix-sentencesegmenter

rename to_flush to concatenate_tsw

move to src

first speaker is "0" no more None

add diarization (beta). Disabled by default

split whisper_online.py into smaller files

del online once webstreaming is finished

specify encoding to ensure Python reads file as UTF-8

no online conflict when multiple users

black formating

Append full transcription in websocket processing

Batch unprocessed audio to reduce Whisper streaming calls

/ws to /asr to distinguish protocol ws:// from endpoint

better buffer gestion

add fastapi server with live webm to pcm conversion and web page showing both complete transcription and partial transcription