VAD causes error

[xevidos]

Which OS are you using?

• OS: Linux Docker

When enabling VAD I get the following error:

Error transcribing file: '<' not supported between instances of 'float' and 'NoneType'
Traceback (most recent call last):
File "/data/.venv/lib/python3.11/site-packages/gradio/queueing.py", line 625, in process_events
response = await route_utils.call_process_api(
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/data/.venv/lib/python3.11/site-packages/gradio/route_utils.py", line 322, in call_process_api
output = await app.get_blocks().process_api(
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/data/.venv/lib/python3.11/site-packages/gradio/blocks.py", line 2047, in process_api
result = await self.call_function(
^^^^^^^^^^^^^^^^^^^^^^^^^
File "/data/.venv/lib/python3.11/site-packages/gradio/blocks.py", line 1594, in call_function
prediction = await anyio.to_thread.run_sync( # type: ignore
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/data/.venv/lib/python3.11/site-packages/anyio/to_thread.py", line 56, in run_sync
return await get_async_backend().run_sync_in_worker_thread(
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/data/.venv/lib/python3.11/site-packages/anyio/_backends/_asyncio.py", line 2505, in run_sync_in_worker_thread
return await future
^^^^^^^^^^^^
File "/data/.venv/lib/python3.11/site-packages/anyio/_backends/_asyncio.py", line 1005, in run
result = context.run(func, *args)
^^^^^^^^^^^^^^^^^^^^^^^^
File "/data/.venv/lib/python3.11/site-packages/gradio/utils.py", line 869, in wrapper
response = f(*args, **kwargs)
^^^^^^^^^^^^^^^^^^
File "/data/config/modules/whisper/base_transcription_pipeline.py", line 229, in transcribe_file
transcribed_segments, time_for_task = self.run(
^^^^^^^^^
File "/data/config/modules/whisper/base_transcription_pipeline.py", line 141, in run
vad_processed, speech_chunks = self.vad.run(
^^^^^^^^^^^^^
File "/data/config/modules/vad/silero_vad.py", line 58, in run
speech_chunks = self.get_speech_timestamps(
^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/data/config/modules/vad/silero_vad.py", line 160, in get_speech_timestamps
if (speech_prob < neg_threshold) and triggered:
^^^^^^^^^^^^^^^^^^^^^^^^^^^
TypeError: '<' not supported between instances of 'float' and 'NoneType'

[xevidos]

After looking into this, it looks to me like the neg_threshold isn't set, perhaps a fallback should be added? Or a default value into the config file?

Old fallback?
if vad_options.neg_threshold is None: vad_options.neg_threshold = threshold - 0.15

[jhj0517]

Hi, Thanks for reporting this! I was able to reproduce the same.
I'm installing faster-whisper from source code because of #424 (comment),

It was probably caused by new source codes updates.
I guess it's time to use the fixed version now, faster-whisper==1.1.1, which was released 2 days ago.

[jhj0517]

Fixed in #441.
Thanks for noticing @xevidos, please feel free to reopen if the problem persists.