Talk video recording transcription (stt_whisper2) can't find produced files.

ā„¹ļø Support šŸ’¬ Talk (spreed)
, , , ,
1
Support intro

Sorry to hear youā€™re facing problems :slightly_frowning_face:

help.nextcloud.com is for home/non-enterprise users. If youā€™re running a business, paid support can be accessed via portal.nextcloud.com where we can ensure your business keeps running smoothly.

In order to help you as quickly as possible, before clicking Create Topic please provide as much of the below as you can. Feel free to use a pastebin service for logs, otherwise either indent short log examples with four spaces:

example

Or for longer, use three backticks above and below the code snippet:

longer
example
here

Some or all of the below information will be requested if it isnā€™t supplied; for fastest response please provide as much as you can :heart:

Some useful links to gather information about your Nextcloud Talk installation:
Information about Signaling server: /index.php/index.php/settings/admin/talk#signaling_server
Information about TURN server: /index.php/settings/admin/talk#turn_server
Information about STUN server: /index.php/settings/admin/talk#stun_server

Nextcloud version: AIONextcloud Hub 26 Spring (35.0.1)
Talk Server version : 2.1.1 docker
Custom Signaling server configured: no
Custom TURN server configured: no
Custom STUN server configured: no

Custom HARP image: not AIOā€™s internal. Separate image (image: Package nextcloud-appapi-harp Ā· GitHub)

exApp added: stt_whisper2 (CPU mode)

Talk>Recording transcription: enabled

--

Dear community!

Iā€™m not sure whether my case is bug or a feature). The feature Iā€™m aiming for is automatic transcription of the video record of the Talk call.

The record itself is saved in the /Records/sth folder in the mp4 format. Quality is good. After the call end, I see the work of the whisper in the containerā€™s log:

TRACE:    HTTP connection made
TRACE:    ASGI [8] Started scope={'type': 'http', 'asgi': {'version': '3.0', 'spec_version': '2.3'}, 'http_version': '1.1', 'server': ('/tmp/exapp.sock', None), 'client': None, 'scheme': 'http', 'root_path': '', 'headers': '<...>', 'state': {}, 'method': 'POST', 'path': '/trigger', 'raw_path': b'/trigger', 'query_string': b'providerId=stt_whisper2%3Afaster-whisper-large-v3'}
TRACE:    ASGI [8] Send {'type': 'http.response.start', 'status': 200, 'headers': '<...>'}
INFO:      - "POST /trigger?providerId=stt_whisper2%3Afaster-whisper-large-v3 HTTP/1.1" 200 OK
TRACE:    ASGI [8] Send {'type': 'http.response.body', 'body': '<2 bytes>'}
TRACE:    ASGI [8] Completed
14:37:10 - stt_whisper2 - INFO - Next task: 4
14:37:10 - stt_whisper2 - INFO - model: faster-whisper-large-v3 enhanced: False
14:37:10 - stt_whisper2 - INFO - generating transcription
14:37:11 - faster_whisper - INFO - Processing audio with duration 09:01.099
14:37:12 - faster_whisper - INFO - VAD filter removed 01:02.107 of audio
TRACE:    HTTP connection lost
14:37:19 - faster_whisper - INFO - Detected language 'ru' with probability 0.99
14:49:44 - stt_whisper2 - INFO - transcription generated: 753.4668653400149s

Also, thereā€™s a CPU load as expected.

However, I canā€™t find a transcription) Where it is supposed to be found? Maybe it is expected way of operation??

If I use AI assistant on the video recording file directly to generate subtitles, the whisper processing repeats, I see the same logs in the stt_whisper2 image. And this way I can save a srt file using a provided ā€˜Save this mediaā€™ action. Thus it is saved at the /Assistant/some-number.srt .

TRACE:    HTTP connection made
TRACE:    ASGI [9] Started scope={'type': 'http', 'asgi': {'version': '3.0', 'spec_version': '2.3'}, 'http_version': '1.1', 'server': ('/tmp/exapp.sock', None), 'client': None, 'scheme': 'http', 'root_path': '', 'headers': '<...>', 'state': {}, 'method': 'POST', 'path': '/trigger', 'raw_path': b'/trigger', 'query_string': b'providerId=stt_whisper2_subtitles%3Afaster-whisper-large-v3'}
TRACE:    ASGI [9] Send {'type': 'http.response.start', 'status': 200, 'headers': '<...>'}
INFO:      - "POST /trigger?providerId=stt_whisper2_subtitles%3Afaster-whisper-large-v3 HTTP/1.1" 200 OK
TRACE:    ASGI [9] Send {'type': 'http.response.body', 'body': '<2 bytes>'}
TRACE:    ASGI [9] Completed
15:59:31 - stt_whisper2 - INFO - Next task: 5
15:59:31 - stt_whisper2 - INFO - model: faster-whisper-large-v3 enhanced: False
15:59:31 - stt_whisper2 - INFO - generating transcription
15:59:32 - faster_whisper - INFO - Processing audio with duration 09:01.099
15:59:33 - faster_whisper - INFO - VAD filter removed 01:02.107 of audio
TRACE:    HTTP connection lost
15:59:40 - faster_whisper - INFO - Detected language 'ru' with probability 0.99
16:11:59 - stt_whisper2 - INFO - transcription generated: 748.1169894807972s

The question is - how to make a file available after the first pass of the transcription?

How is this whole transcription supposed to work?

Iā€™ve installed live transcription also, it works, but it doesnā€™t produce any file output, so it is not used anymore.