Cannot install deepspeech of python

Question

I want to use DeepSpeech of Mozilla on my Linux 22.04 system, following this website:

https://deepspeech.readthedocs.io/en/r0.9/?badge=latest 

At the very beginning line, at

pip3 install deepspeech

I got this error:

ERROR: Could not find a version that satisfies the requirement deepspeech (from versions: none)
ERROR: No matching distribution found for deepspeech

I ran into internet and followed all methods such as upgrading pip3, using pip instead of pip3. I could not solve the problem.

This website:

https://github.com/mozilla/DeepSpeech/issues/3693

suggests to use archive. I did not understand which repository should I archive at this step.

It is very nice of you if you can help me.

Answer 1

It appears that DeepSpeech is no longer maintained although there is no explicit message to users. However, both current and new users are continuing to submit pull requests and open issues with the assumption that the project is still actively being developed.

Check this link: https://github.com/mozilla/DeepSpeech/issues/3693

Answer 2

try this uninstall current python on your pc. Then, install python version 3.7.3 and then again try pip install deepspeech. This really works!

Answer 3

in 2021

https://github.com/mozilla/DeepSpeech

became

https://github.com/coqui-ai/STT

as can be seen from the commit history, but is recommending

https://github.com/openai/whisper

Answer 4

Got it working using a python3.7 virtualenv. It seems that python 3.7 is the last supported version for using deepspeech.

I'm in arch linux, so first I had to install python3.7 from the AUR:

yay -S python37

Then, inside my project folder, I created a venv:

virtualenv -p /usr/bin/python3.7 venv

After that, activate the venv:

source ./venv/bin/activate

Then, you can safely do

pip install deepspeech

Answer 5

I need to develop a transcriber tool using Python, so I did some research, and this was the result:

• DeepSpeech - No longer maintained
• Coqui - No longer actively maintained
• Whisper - Actively maintained - Applies spelling corrections to the transcription
• Vosk - Actively maintained (uses Kaldi as core) - transcribe it in the same way it was spoken

I did a couple of tests and got great accuracy vs performance using both Whisper and Vosk. BTW, I noticed a slight difference in their transcription results. For example, I have an audio that the speaker says I'm gonna get, so Vosk transcribed it to I'm gonna get, and Whisper transcribed to I am going to get.