Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

从15秒开始之后的文字全部相同 #23

Closed
sandycs-protoss opened this issue Sep 15, 2023 · 10 comments
Closed

从15秒开始之后的文字全部相同 #23

sandycs-protoss opened this issue Sep 15, 2023 · 10 comments
Labels
enhancement New feature or request

Comments

@sandycs-protoss
Copy link

用的是largev1模型。
这个里不能上图,如链接所示:
https://storage.googleapis.com/taotio/memobug.jpg

具体表现是英文视频转换完成后,从第15秒开始内容全部相同

@sandycs-protoss sandycs-protoss added the bug Something isn't working label Sep 15, 2023
@Makememo
Copy link
Owner

请先选择视频源语言,再右上角翻译。

@sdugoten
Copy link

sdugoten commented Sep 27, 2023

預到一樣的問題, 前面在5分鐘左右出現過一次, 使用Large v2 model. 到最後面10分鐘左右又是這樣, 全部一樣的內容, 你可以download 原檔試.

https://temp-file.org/ANOxf418CwZdAwe/file

我已經先選日本語為"视频源语言"

ogfb2V

ogfb2V

直接用 whisper 出來的檔沒事, 一樣是Large v2 model. 所以應該是 Memo 內的問題

ogfb2V

另外, 我嘗試再encode 上面的原片, 用較高的audio encoding rate. 結果....memo 只聽到音樂..

ogfb2V

@Makememo
Copy link
Owner

2023-09-27 at 10 10 35@2x
我加了提示词后正常了。

@sdugoten
Copy link

提示词

請問你用甚麼提示词?

@sdugoten
Copy link

2023-09-27 at 10 10 35@2x 我加了提示词后正常了。

我想我找到解決方案. 請問你可不可以加 --temperature_increment_on_fallback , --condition_on_previous_text 這2個switch? 因為解決方案在這裡

https://blog.gdeltproject.org/experiments-with-whisper-asr-model-parameters-non-determinism-temperature_increment_on_fallback/

https://github.com/openai/whisper/pull/1253

應該加了這2個switch 後, 我會直接付費買lifetime 版. 謝啦.

@Makememo
Copy link
Owner

OK,后续会把参数开放开来。

@sdugoten
Copy link

OK,后续会把参数开放开来。

另外, 好像提示詞跟whisper 有點不一樣. 我在whisper 加入這句 "以下係香港嘅港式廣東話", 片段內容是廣東話的話, 會直接輸出廣東話, 不過在memo 用一樣的提示詞, 輸出還是會翻譯成正體書面語.

@Makememo Makememo added enhancement New feature or request and removed bug Something isn't working labels Oct 21, 2023
@Makememo
Copy link
Owner

Makememo commented Nov 4, 2023

已经上线。https://memo.ac/releases.html

@Makememo Makememo closed this as completed Nov 4, 2023
@sdugoten
Copy link

sdugoten commented Nov 4, 2023

  1. --temperature_increment_on_fallback , --condition_on_previous_text 這2個switch在那找到? 人声检测提取(实验功能)對嗎?
  2. 提示詞輸入 "以下係香港嘅港式廣東話" 仍然不會顯視video 內的廣東話, 直接翻譯成中文

@sdugoten
Copy link

sdugoten commented Nov 4, 2023

看來人声检测提取(实验功能)好像解決我之前給你們那個video 的問題. 我另外再開一個新的issue 關於上面point 2 的提示詞問題

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
enhancement New feature or request
Projects
None yet
Development

No branches or pull requests

3 participants