Not working properly on CachyOS Linux.
Whenever I run the program and try to speak the result always says "You."
ALSA lib pcm.c:2722:(snd_pcm_open_noupdate) Unknown PCM cards.pcm.rearALSA lib pcm.c:2722:(snd_pcm_open_noupdate) Unknown PCM cards.pcm.center_lfeALSA lib pcm.c:2722:(snd_pcm_open_noupdate) Unknown PCM cards.pcm.sideALSA lib pcm_route.c:878:(find_matching_chmap) Found no matching channel mapALSA lib pcm_route.c:878:(find_matching_chmap) Found no matching channel mapALSA lib pcm_route.c:878:(find_matching_chmap) Found no matching channel mapALSA lib pcm_route.c:878:(find_matching_chmap) Found no matching channel mapCannot connect to server socket err = No such file or directoryCannot connect to server request channeljack server is not running or cannot be startedJackShmReadWritePtr::~JackShmReadWritePtr - Init not done for -1, skipping unlockJackShmReadWritePtr::~JackShmReadWritePtr - Init not done for -1, skipping unlockCannot connect to server socket err = No such file or directoryCannot connect to server request channeljack server is not running or cannot be startedJackShmReadWritePtr::~JackShmReadWritePtr - Init not done for -1, skipping unlockJackShmReadWritePtr::~JackShmReadWritePtr - Init not done for -1, skipping unlockALSA lib pcm_oss.c:404:(_snd_pcm_oss_open) Cannot open device /dev/dspALSA lib pcm_oss.c:404:(_snd_pcm_oss_open) Cannot open device /dev/dspALSA lib pcm_a52.c:1036:(_snd_pcm_a52_open) a52 is only for playbackALSA lib confmisc.c:160:(snd_config_get_card) Invalid field cardALSA lib pcm_usb_stream.c:481:(_snd_pcm_usb_stream_open) Invalid card 'card'ALSA lib confmisc.c:160:(snd_config_get_card) Invalid field cardALSA lib pcm_usb_stream.c:481:(_snd_pcm_usb_stream_open) Invalid card 'card'Cannot connect to server socket err = No such file or directoryCannot connect to server request channeljack server is not running or cannot be startedJackShmReadWritePtr::~JackShmReadWritePtr - Init not done for -1, skipping unlockJackShmReadWritePtr::~JackShmReadWritePtr - Init not done for -1, skipping unlock
I'm not really that knowledgeable, but from asking ChatGPT the errors you see are probably related to the audio handling of the OS and not RealtimeSTT. You can see the full answer here if interested.
Whenever I run the program and try to speak the result always says "You."
I'm pretty sure that's a whisper hallucination when trying to transcribe silence, which reinforces ChatGPT's diagnosis.
You might also want to supply some more info about your hardware and software configuration. Especially:
- python version
- CUDA Toolkit version (if applicable)
- GPU model (if applicable)
- pytorch version & configuration (CPU/CUDA XX.X/ROCm 6.2)
- whether you're using conda or pip
- which script of the scripts you're using and how you have it configured.
is CUDA Toolkit really required to use cuda? I mean, pytorch seems work fine without installing these toolkit on my system, so I just wondering:
$ python
Python 3.12.7 (main, Oct 1 2024, 11:15:50) [GCC 14.2.1 20240910] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import torch
>>> torch.cuda.is_available()
True
>>> torch.cuda.device_count()
1
>>> torch.cuda.current_device()
0
>>> torch.cuda.get_device_name(0)
'NVIDIA GeForce RTX 4070 Laptop GPU'
Oh nvm, I think it required because the program need a cuda shared library, ~so I ended up install it from arch package repository anyway~.
EDIT:
but wait, the nvidia python package already have it?? found it on /home/RealtimeSTT/.venv/lib/python3.12/site-packages/nvidia/cudnn/lib/libcudnn.so.9, so why the program can't found it? Unable to load any of {libcudnn_ops.so.9.1.0, libcudnn_ops.so.9.1, libcudnn_ops.so.9, libcudnn_ops.so}
EDIT:
ah, turns out I need to specify the path on LD_LIBRARY_PATH