Skip to content

Add VAD pre-filter for transcription #69

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
marcelklehr merged 3 commits into from
May 11, 2026
Merged

Add VAD pre-filter for transcription #69

Changes from 1 commit
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
91f0c1e
Add Silero VAD pre-filter for transcription
nilsimda Apr 26, 2026
42bf6d3
fix: Add a nice error message for parse errors
marcelklehr May 11, 2026
3e7a4c8
fix: Add VAD filter env vars to info.xml
marcelklehr May 11, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
17 changes: 16 additions & 1 deletion lib/main.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -43,6 +43,17 @@
LOGGER = logging.getLogger(os.environ["APP_ID"])
LOGGER.setLevel(logging.DEBUG)

# --- VAD configuration (Silero, bundled with faster-whisper) ---
# Enabled by default; disable by setting STT_WHISPER2_VAD_FILTER=0 in the container env.
# Tunables map 1:1 onto faster_whisper.vad.VadOptions.
VAD_FILTER = os.environ.get("STT_WHISPER2_VAD_FILTER", "1") == "1"
VAD_PARAMETERS = {
Comment thread
marcelklehr marked this conversation as resolved.
Outdated
"threshold": float(os.environ.get("STT_WHISPER2_VAD_THRESHOLD", "0.5")),
"min_speech_duration_ms": int(os.environ.get("STT_WHISPER2_VAD_MIN_SPEECH_MS", "0")),
"min_silence_duration_ms": int(os.environ.get("STT_WHISPER2_VAD_MIN_SILENCE_MS", "2000")),
"speech_pad_ms": int(os.environ.get("STT_WHISPER2_VAD_SPEECH_PAD_MS", "400")),
}


Comment thread
marcelklehr marked this conversation as resolved.
Outdated
def load_models():
models = {}
Expand Down Expand Up @@ -144,7 +155,11 @@ def background_thread_task():
LOGGER.info("generating transcription")
time_start = perf_counter()
file_name = get_file(nc, task["id"], task.get("input").get('input'))
segments, info = model.transcribe(file_name)
segments, info = model.transcribe(
file_name,
vad_filter=VAD_FILTER,
vad_parameters=VAD_PARAMETERS if VAD_FILTER else None,
)
transcript = ''
for segment in segments:
transcript += segment.text
Expand Down
Loading