Failed to load audio: Cannot open shim file for read

[daytodatainc]

I am unable to transcribe the audio to text using Jupyter. It works perfectly on Colab, but on my local machine it doesn't.
Code:
audio_files = []
for (r,d,f) in os.walk(audio_dir):
audio_files.append(f) #append the audio file name to the array

audio_files = audio_files[0]
print(f"There are {len(audio_files)} audio file(s) loaded")

for audio_file in audio_files:
result = model.transcribe(r"./audio_files/"+audio_file, fp16=False, language="en")
print(f"{audio_file} has been converted")
print(len(result['text']))

if len(result['text']) > 0:
split_values = audio_file.replace("CXone recording_","").replace("[UTC]","").split("_")[:3]
date_value = split_values[1]
time_value = split_values[2].replace("-",":")
person_value = split_values[0]
file_name = f"Type: {model_type} {date_value} - {time_value} - {person_value}.txt"
print(file_name)
full_file_name = f"./audio_files/audio_text/{file_name}"

text_file = open(full_file_name, "w")
n = text_file.write(result['text'])
text_file.close()
print(f"The file has been written: {file_name}")`

Error:

Error Traceback (most recent call last)
File c:\users\cmw1s\appdata\local\programs\python\python38\lib\site-packages\whisper\audio.py:42, in load_audio(file, sr)
38 try:
39 # This launches a subprocess to decode audio while down-mixing and resampling as necessary.
40 # Requires the ffmpeg CLI and ffmpeg-python package to be installed.
41 out, _ = (
---> 42 ffmpeg.input(file, threads=0)
43 .output("-", format="s16le", acodec="pcm_s16le", ac=1, ar=sr)
44 .run(cmd=["ffmpeg", "-nostdin"], capture_stdout=True, capture_stderr=True)
45 )
46 except ffmpeg.Error as e:

File c:\users\cmw1s\appdata\local\programs\python\python38\lib\site-packages\ffmpeg_run.py:325, in run(stream_spec, cmd, capture_stdout, capture_stderr, input, quiet, overwrite_output)
324 if retcode:
--> 325 raise Error('ffmpeg', out, err)
326 return out, err

Error: ffmpeg error (see stderr output for detail)

The above exception was the direct cause of the following exception:

RuntimeError Traceback (most recent call last)
Cell In [4], line 14
9 #1. convert the audio file to text using Whisper
10 #2. create the text file
11 #3. save the text file in the audio_text directory
12 for audio_file in audio_files:
13 # convert the audio file to text
---> 14 result = model.transcribe(r"./audio_files/"+audio_file, fp16=False, language="en")
15 print(f"{audio_file} has been converted")
16 print(len(result['text']))

File c:\users\cmw1s\appdata\local\programs\python\python38\lib\site-packages\whisper\transcribe.py:84, in transcribe(model, audio, verbose, temperature, compression_ratio_threshold, logprob_threshold, no_speech_threshold, condition_on_previous_text, **decode_options)
81 if dtype == torch.float32:
82 decode_options["fp16"] = False
---> 84 mel = log_mel_spectrogram(audio)
86 if decode_options.get("language", None) is None:
87 if not model.is_multilingual:

File c:\users\cmw1s\appdata\local\programs\python\python38\lib\site-packages\whisper\audio.py:111, in log_mel_spectrogram(audio, n_mels)
109 if not torch.is_tensor(audio):
110 if isinstance(audio, str):
--> 111 audio = load_audio(audio)
112 audio = torch.from_numpy(audio)
114 window = torch.hann_window(N_FFT).to(audio.device)

File c:\users\cmw1s\appdata\local\programs\python\python38\lib\site-packages\whisper\audio.py:47, in load_audio(file, sr)
41 out, _ = (
42 ffmpeg.input(file, threads=0)
43 .output("-", format="s16le", acodec="pcm_s16le", ac=1, ar=sr)
44 .run(cmd=["ffmpeg", "-nostdin"], capture_stdout=True, capture_stderr=True)
45 )
46 except ffmpeg.Error as e:
---> 47 raise RuntimeError(f"Failed to load audio: {e.stderr.decode()}") from e
49 return np.frombuffer(out, np.int16).flatten().astype(np.float32) / 32768.0

RuntimeError: Failed to load audio: Cannot open shim file for read.
Could not read shim file.

[phineas-pta]

u should read error message more throughly

Requires the ffmpeg CLI and ffmpeg-python package to be installed.

[daytodatainc]

Thanks for the snarky response. I followed the git-hub requirements on making sure I've installed the right packages.

C:>pip list
Package Version

---

certifi 2023.5.7
charset-normalizer 3.1.0
colorama 0.4.6
ffmpeg-python 0.2.0
filelock 3.12.2

When I run this on Google Colab, no issues but locally I run into these issues.

[phineas-pta]

did u install ffmpeg ?