Help with debugging incorrect speech timestamps

[sha-roze]

Hi, I'm having issues with getting the timestamps of this video.

Here is my code and output:

import torch

torch.set_num_threads(1)

model, utils = torch.hub.load(repo_or_dir="snakers4/silero-vad", model="silero_vad")
(get_speech_timestamps, _, read_audio, _, _) = utils

wav = read_audio(
    r"C:\Users\Sha Roze\Downloads\Videos\httpswww.youtube.comwatchv=oM5hNuAmWs0.wav"
)
speech_timestamps = get_speech_timestamps(
    wav,
    model,
    return_seconds=True,  # Return speech timestamps in seconds (default is samples)
)

print(speech_timestamps)

Using cache found in C:\Users\Sha Roze/.cache\torch\hub\snakers4_silero-vad_master
[{'start': 3.2, 'end': 4.5}]

[snakers4]

v5 has problems with singing, but v3.1 works:

  model, utils = torch.hub.load(repo_or_dir='snakers4/silero-vad:v3.1',
                                model='silero_vad',
                                force_reload=True,
                                onnx=USE_ONNX)

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

wav = read_audio('prayer.wav', sampling_rate=SAMPLING_RATE)
# get speech timestamps from full audio file
speech_timestamps = get_speech_timestamps(wav, model,
                                          sampling_rate=SAMPLING_RATE,
                                          visualize_probs=True,
                                          threshold=0.1, return_seconds=True)
pprint(speech_timestamps)

[sha-roze]

Thank you for answering, this worked for me.