docker env[DEBUG] not work

[Aliang-code]

I can already see the DEBUG variable I set in the container, but still don't have any debug log output when I post request /jellyfin because it didn't start when the media is automatically added, with env PROCADDEDMEDIA=False SKIPIFINTERNALSUBLANG=chi SKIPIFEXTERNALSUB=True, so i need see the debug logs

root@fd091c0a463d:/subgen# echo $DEBUG
True

subgen.env file not found. Please run prompt_and_save_env_variables() first.

subgen.py exists and UPDATE is set to False, skipping download.

Launching subgen.py

File subgen.env not found. Environment variables not set.

File subgen.env not found. Environment variables not set.

INFO:root:Subgen v2024.7.13.88

INFO:root:Starting Subgen with listening webhooks!

INFO:root:Transcriptions are limited to running 2 at a time

INFO:root:Running 4 threads per transcription

INFO:root:Using cpu to encode

INFO:root:Using faster-whisper

�[32mINFO�[0m:     Started server process [�[36m27�[0m]

�[32mINFO�[0m:     Waiting for application startup.

�[32mINFO�[0m:     Application startup complete.

�[32mINFO�[0m:     Uvicorn running on �[1mhttp://0.0.0.0:9000�[0m (Press CTRL+C to quit)

�[32mINFO�[0m:     192.168.1.251:55520 - "�[1mPOST /jellyfin HTTP/1.1�[0m" �[32m200 OK�[0m

[McCloudS]

You would want PROCADDEDMEDIA=False to be true if you want it to run on a file being added to the library.

[Aliang-code]

You would want PROCADDEDMEDIA=False to be true if you want it to run on a file being added to the library.

yes,I informed its usage after checking the source code, but the description in the readme is a little confusing.

[Aliang-code]

You would want PROCADDEDMEDIA=False to be true if you want it to run on a file being added to the library.

After a day of intensive use, I think I'm close to my goal: generating Chinese subtitles for those English videos where subtitles cannot be found.
Even though I know whisper doesn't officially support translation to other languages,but it does work.
The only problem is that after a long period of transcription, once multiple English words appear in the context (such as name), the subsequent output will all become English.
However, the built-in silero-vad of faster_whisper solved this problem to a certain extent after splitting the fragments, and then adjusted the prompt_reset_on_temperature parameter. Now it works very well for me, so can you provide these two parameters to the env configuration?

model.transcribe_stable(file_path, vad=True, prompt_reset_on_temperature=0.35, language=force_language, task=transcription_type, progress_callback=progress, initial_prompt=custom_model_prompt, regroup=custom_regroup)

My transcription result example for Dinosaurs (1991) S1E4

• default to vad=False, prompt_reset_on_temperature=0.5

[00:03:39,500 --> 00:03:42,160]	他面前有雕塑
[00:03:42,160 --> 00:03:43,328]	Gary
[00:03:44,096 --> 00:03:45,888]	他的名字是Gary
[00:03:46,350 --> 00:03:48,224]	Terror有个新的名字
[00:03:48,560 --> 00:03:49,376]	Gary
[00:03:50,140 --> 00:03:51,280]	这是个好地方
[00:03:53,024 --> 00:03:54,160]	从法律部长
[00:03:54,160 --> 00:03:55,700]	Spikeback
[00:03:55,700 --> 00:03:56,580]	Ribcrusher
[00:03:56,580 --> 00:03:57,440]	Fenske
[00:03:58,560 --> 00:03:59,840]	Dear Mr. Sinclair
[00:03:59,840 --> 00:04:00,860]	I'm writing to you
[00:04:00,860 --> 00:04:02,800]	on behalf of my client Gary
[00:04:02,800 --> 00:04:04,160]	who met your wife
[00:04:04,160 --> 00:04:05,100]	at the supermarket
[00:04:05,100 --> 00:04:06,600]	and who would like to take her
[00:04:06,600 --> 00:04:07,660]	as his mate
[00:04:07,660 --> 00:04:09,860]	under section 412
[00:04:09,860 --> 00:04:10,740]	article 9
[00:04:10,740 --> 00:04:11,990]	of the code of the wilderness
... After that it’s all in English

• after set vad=True, prompt_reset_on_temperature=0.35

[00:03:39,480 --> 00:03:42,160]	他面前有雕像
[00:03:42,160 --> 00:03:43,328]	Gary
[00:03:44,096 --> 00:03:45,856]	他的名字是Gary
[00:03:46,880 --> 00:03:48,192]	Terror有个新的名字
[00:03:48,900 --> 00:03:49,376]	Gary
[00:03:50,016 --> 00:03:51,400]	那是个好地方
[00:03:53,024 --> 00:03:54,160]	从法律部队
[00:03:54,160 --> 00:03:55,700]	有Spikeback
[00:03:55,700 --> 00:03:56,580]	Ribcrusher
[00:03:56,580 --> 00:03:57,440]	和Fenske
[00:03:58,560 --> 00:03:59,900]	亲爱的Mr. Sinclair
[00:04:00,032 --> 00:04:02,200]	在我客户面前
[00:04:02,200 --> 00:04:02,816]	Gary
[00:04:03,040 --> 00:04:05,120]	他在市场上
[00:04:05,120 --> 00:04:06,600]	他想要把她
[00:04:06,600 --> 00:04:08,032]	当他妻子
[00:04:08,352 --> 00:04:09,740]	在412条
[00:04:09,740 --> 00:04:10,700]	第九章
[00:04:10,700 --> 00:04:12,096]	《地狱文章》
... no longer English here

[McCloudS]

I haven't had time to extensively test this yet, but give it a shot. I added a SUBGEN_KWARGS env variable. You should be able to override or add most options.

It’s a simple kwarg dictionary.

So if you set it to: {'vad': 'True','prompt_reset_on_temperature': '0.35'} it ought to get you close. I tested loosely on unraid and it didn't throw errors outright.

I don't want to document it until it's known to be working.

[Aliang-code]

I haven't had time to extensively test this yet, but give it a shot. I added a SUBGEN_KWARGS env variable. You should be able to override or add most options.

It’s a simple kwarg dictionary.

So if you set it to: {'vad': 'True','prompt_reset_on_temperature': '0.35'} it ought to get you close. I tested loosely on unraid and it didn't throw errors outright.

I don't want to document it until it's known to be working.

thx, it works well !

environment:
  - "SUBGEN_KWARGS={'vad': True,'prompt_reset_on_temperature': 0.35}"