Couldn't find appropriate backend to handle uri /tmp/en_example.wav and format None.

[escabar]

Trying to dockerize silero-vad for AWS Lambda and running into the following.
Note: I'm using the example from here.

error message:

{"errorMessage": "Couldn't find appropriate backend to handle uri /tmp/en_example.wav and format None.", "errorType": "RuntimeError", "requestId": "d324dc74-ae5b-4719-b3e7-4499e160a517", "stackTrace": ["  File \"/var/task/lambda_function.py\", line 116, in handler\n    wav = read_audio('/tmp/en_example.wav', 16000)\n", "  File \"/root/.cache/torch/hub/snakers4_silero-vad_master/utils_vad.py\", line 136, in read_audio\n    wav, sr = torchaudio.load(path)\n", "  File \"/var/lang/lib/python3.12/site-packages/torchaudio/_backend/utils.py\", line 204, in load\n    backend = dispatcher(uri, format, backend)\n", "  File \"/var/lang/lib/python3.12/site-packages/torchaudio/_backend/utils.py\", line 116, in dispatcher\n    raise RuntimeError(f\"Couldn't find appropriate backend to handle uri {uri} and format {format}.\")\n"]}%

Dockerfile:

FROM public.ecr.aws/lambda/python:3.12

# Copy requirements.txt
COPY requirements.txt ${LAMBDA_TASK_ROOT}

# Install the specified packages
RUN pip install -r requirements.txt

# Copy function code
COPY lambda_function.py ${LAMBDA_TASK_ROOT}
COPY bin/ffmpeg /bin

# Set the CMD to your handler (could also be done as a parameter override outside of the Dockerfile)
CMD [ "lambda_function.handler" ]

requirements.txt :

torch
torchaudio
ffmpeg-python
numpy
soundfile
sox

Lambda_function:

import torchaudio
import torch

SAMPLING_RATE = 16000
torch.set_num_threads(1)

def handler(event, context):
    torch.hub.download_url_to_file('https://models.silero.ai/vad_models/en.wav', 'en_example.wav')
    
    model, utils = torch.hub.load(repo_or_dir='snakers4/silero-vad',
                                model='silero_vad',
                                force_reload=True,
                                onnx=False)

    (get_speech_timestamps,
    save_audio,
    read_audio,
    VADIterator,
    collect_chunks) = utils

    wav = read_audio('en_example.wav', sampling_rate=SAMPLING_RATE)
    # get speech timestamps from full audio file
    speech_timestamps = get_speech_timestamps(wav, model, sampling_rate=SAMPLING_RATE)
    print(speech_timestamps)

Any suggestions?

[snakers4]

Hi,

This thread may be helpful - facebookresearch/demucs#570

Note: I'm using the example from here.

Also please check out the examples from the repo, the torch.hub example was not updated for quite some time.

[escabar]

Thank you. Appears that I needed sox, ffmpeg and soundfile all installed to get it working.