Comments (3)
Is what I am looking for possible?
Having a set of hypothesis for the words makes sense.
But having timestamps for all the alternatives is not really possible.
Also the confidence score should be different for all alternatives.
And this is more a request for openai-whisper, not whisper-timestamped.
from whisper-timestamped.
I would be happy with having multiple hypotheses for the words, and using the timestamp of the top candidate. Also I don't need the confidence scores for my voting algorithm that I am implementing.
Do you think you could at least provide as output N hypotheses in full, instead of word-by-word N hypotheses? Or is this also an openai-whisper request? I just don't know how I would then proceed to extract timestamps for each word in each full hypothesis.
from whisper-timestamped.
I managed to modify the openai-whisper transcribe.py
and decoding.py
files to return multiple hypotheses to whisper-timestamped transcribe.py
. From there, I was able to extract word-level time alignments for each hypotheses by modifying some of the code. I can provide more details if anyone needs to extract word-level time alignments for multiple hypotheses.
from whisper-timestamped.
Related Issues (20)
- Broken link for plotting word alignment section HOT 7
- Loading finetuned model serialized with safetensors (and/or sharded models) HOT 10
- How to activate flash attention? HOT 2
- Could it be possible to apply the same technique to the whisper API? HOT 6
- ctranslate2 support HOT 1
- CPU only light install links are broken? HOT 3
- Issue with accented characters coming up as symbols in output json file
- Repetitive Phrase Looping HOT 3
- Bad timestamp prediction with some finetuned Whisper models HOT 9
- How to use cuda? HOT 2
- cuda is not available?
- cuda is not available? HOT 8
- There are plans to use ctranslate2 to speed up? similar to faster-whisper HOT 2
- Whether --max_line_width and --max_line_count are not supported? HOT 2
- When the transcription progress reaches 100%, it takes a long time to show the result, sometimes even up to 10 minutes. HOT 1
- What is it (very slowly) trying to download? HOT 4
- Output filenames aren't consistent with original openai-whisper implementation HOT 1
- Output syntaxically invalid vtt file (header appears twice) HOT 1
- Incorrect timestamps when using VAD with large model only
- Timestamps not provided per-word for Chinese/Japanese/Korean HOT 3
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from whisper-timestamped.