slesinger / czech-speech-to-text

czech language, stt, speech to text, speech recognition, voice transcription, asr, offline, free
Apache License 2.0
20 stars 2 forks source link

docker-based server crashes upon any WAV file upload #4

Open mmokrejs opened 2 weeks ago

mmokrejs commented 2 weeks ago

Hi, I am crashing the server when I upload the data. I thought I have to convert from 48kHz/stereo to 16kHz/mono but even after doing that I still get crashes.

$ sudo docker run -v `pwd`:/workspace/data --tmpfs /tmp -p 8888:8888 -p 10456:10456 --net=host --ipc=host --name deepspeech -e MODEL=/workspace/data/deepspeech_final.pth -e LM=/workspace/data/lm.bin slesinger/deepspeech:latest
WARNING: Published ports are discarded when using host network mode
Model  $MODEL: /workspace/data/deepspeech_final.pth
LM binary $LM: /workspace/data/lm.bin
/opt/conda/lib/python3.7/site-packages/omegaconf/omegaconf.py:647: UserWarning: update() merge flag is is not specified, defaulting to False.
For more details, see https://github.com/omry/omegaconf/issues/367
  stacklevel=1,
[2024-10-28 17:37:04,077][root][INFO] - Setting up server...
[2024-10-28 17:39:56,629][root][INFO] - Server initialised
 * Serving Flask app "server" (lazy loading)
 * Environment: production
   WARNING: This is a development server. Do not use it in a production deployment.
   Use a production WSGI server instead.
 * Debug mode: on
[2024-10-28 17:39:58,066][werkzeug][INFO] -  * Running on http://0.0.0.0:10456/ (Press CTRL+C to quit)
[2024-10-28 17:40:04,381][root][INFO] - Transcribing file...
/workspace/deepspeech.pytorch/server.sh: line 6:     8 Segmentation fault      python server.py host=0.0.0.0 port=10456 model.use_half=false model.cuda=false model.model_path=$MODEL lm.decoder_type=beam lm.lm_path=$LM
$
$ ls -la
total 1342380
-rw-r--r-- 1 mmokrejs mmokrejs  174043992 Oct 28 17:53 deepspeech_final.pth
-rw-r--r-- 1 mmokrejs mmokrejs 1200207755 Oct 28 17:56 lm.bin
...