Bug 441020 - speech to text module does not work for me
Summary: speech to text module does not work for me
Status: RESOLVED FIXED
Alias: None
Product: kdenlive
Classification: Applications
Component: User Interface & Miscellaneous (show other bugs)
Version: 21.04.3
Platform: Appimage Linux
: NOR normal
Target Milestone: ---
Assignee: Jean-Baptiste Mardelle
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2021-08-15 22:15 UTC by turtle.engr
Modified: 2022-01-13 18:19 UTC (History)
3 users (show)

See Also:
Latest Commit:
Version Fixed In:
Sentry Crash Report:
fritzibaby: timeline_corruption+


Attachments
stdout and stderr output from running kdenlive test. (20.09 KB, text/x-log)
2021-08-15 22:15 UTC, turtle.engr
Details

Note You need to log in before you can comment on or make changes to this bug.
Description turtle.engr 2021-08-15 22:15:52 UTC
Created attachment 140757 [details]
stdout and stderr output from running kdenlive test.

SUMMARY

I followed the install and usage directions. No subtitle text is created.
Looking at Bug 439757, I made sure vosk is version 0.3.30.

I tried with these appimages:
    kdenlive-21.04.3b-x86_64.appimage
    kdenlive-21.11.70-9295653-x86_64.appimage

--------------------
STEPS TO REPRODUCE

Install steps
1. Followed the steps for installing the python software.
2. Followed the steps for installing the models. Installed in "custom" dir:
  ~/Downloads/vosk-model-small-en-us-0.15/
3 Clicked on "Check configuration" button
  Saw "Speech to text is properly configured."

Apparently, the configuration check is not checking enough, because
speech2text didn't work.

Steps to reproduce
1. Started a new project: speech2text-test.kdenlive
2. Added a video clip with some spoken English audio
3. Defined a short 10sec in/out zone on the timeline
4. Selected the "Edit Subtitle Tool" icon
5. In the Subtitles track, selected the "Speech recognition" icon
6. Selected "Timeline zone (all tracks)", and vosk-model-small-en-us-0.15
7. Clicked on "Process"

--------------------
OBSERVED RESULT
* Saw a progress bar for less than a second.
* Then, "Subtitles imported"
* There was no title text in the subtitle track
* An *empty* speech2text-test.kdenlive.srt file was created

--------------------
EXPECTED RESULT
* I expected to see some subtitle text in the track and in the srt file.

--------------------
SOFTWARE/OS VERSIONS

Linux/KUbuntu
KDE Plasma Version: 5.3.0
KDE Frameworks: 5.44.0-0ubuntu3
Qt Version: 5.9.5

Details:

> cat /etc/os-release
    NAME="Ubuntu"
    VERSION="18.04.5 LTS (Bionic Beaver)"
    ID_LIKE=debian
    PRETTY_NAME="Ubuntu 18.04.5 LTS"
    VERSION_ID="18.04"
    VERSION_CODENAME=bionic

> uname -a
    Linux plasma 5.3.0-7648-generic #41~1586790036~18.04~600aeb5~dev-Ubuntu SMP Mon Apr 13 17:49:24  x86_64 x86_64 x86_64 GNU/Linux

GUI and other packages
    ii kubuntu-desktop 1.370 amd64 Kubuntu Plasma Desktop/Netbook system
    ii plasma-desktop 4:5.12.9.1-0ubuntu0.1 amd64 Tools and widgets for
        the desktop
    ii gdm3 3.28.3-0ubuntu18.04.5pop1~1551986836~18.04~d21ecd7~dev amd64
        GNOME Display Manager
    ii qt5-default:amd64 5.9.5+dfsg-0ubuntu2.5 amd64 Qt 5 development
       defaults package
    ii ffmpeg 7:3.4.8-0ubu amd64 Tools for transcoding, streaming

> python --version
    Python 2.7.17

> pip install --upgrade vosk
    Requirement already satisfied: vosk in /usr/local/lib/python3.6/dist-packages (0.3.30)
    Requirement already satisfied: cffi>=1.0 in /usr/local/lib/python3.6/dist-packages (from vosk) (1.14.6)
    Requirement already satisfied: pycparser in /usr/local/lib/python3.6/dist-packages (from cffi>=1.0->vosk) (2.20)

> /etc# pip install --upgrade srt
    Requirement already satisfied: srt in /usr/local/lib/python3.6/dist-packages (3.4.1)

kdenlive versions tried:
    kdenlive-21.04.3b-x86_64.appimage
    kdenlive-21.11.70-9295653-x86_64.appimage

I would have tried kdenlive-21.08, but I couldn't find a download link for it.
Comment 1 emohr 2021-08-26 19:27:19 UTC
Please try with AppImage version 21.08.0 from the download page: https://kdenlive.org/en/download/. I tested speech to text and it should work.
Comment 2 turtle.engr 2021-08-26 22:25:19 UTC
I tried again with: kdenlive-21.08.0a-x86_64.appimage
Same result. No subtitles and the kdenlive.srt file is empty.

I'll see about digging up an old computer, do a fresh OS install, then try with that. If it works, then I can start looking for the differences between the computers.
Comment 3 Julius Künzel 2022-01-13 18:19:33 UTC
Thanks for your report!

There was an issue with the VOSK api that is fixed in version 21.12 so I am closing this bug know. If you still experience this issue with the latest version feel free to reopen this report and update the software version.