# Help about Whisper chunk\_length

**URL:** <https://discuss.huggingface.co/t/help-about-whisper-chunk-length/140892>\
**Category:** Beginners\
**Created:** [February 14, 2025, 5:27pm UTC](https://discuss.huggingface.co/t/help-about-whisper-chunk-length/140892 "2025-02-14T17:27:25Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![asmacats](https://avatars.discourse-cdn.com/v4/letter/a/e480ec/32.png) [@asmacats](https://discuss.huggingface.co/u/asmacats)\
**Post date:** [February 14, 2025, 5:27pm UTC](https://discuss.huggingface.co/t/help-about-whisper-chunk-length/140892/1 "2025-02-14T17:27:25Z")

</div>

Hello,

I need your help for clarifying why Whisper chunked algorithm should be used when:

1. Transcription speed is the most important factor
2. You are transcribing a **single** long audio file.

I try to use it with different batches and different audio in parallel and it works fine but i have remarked that inference time is huge compared with given audios one by one. I am using A100 with 80 GO, 10 CPU and 60 Go of memory.

Any explanation please ?

---

<div class="post-metadata">

**Author:** ![John6666](https://sea2.discourse-cdn.com/hellohellohello/user_avatar/discuss.huggingface.co/john6666/32/27664_2.png) [@John6666](https://discuss.huggingface.co/u/John6666)\
**Post date:** [February 15, 2025, 7:43am UTC](https://discuss.huggingface.co/t/help-about-whisper-chunk-length/140892/2 "2025-02-15T07:43:53Z")

</div>

Whisper is excellent, but it’s also stubborn. It’s difficult to make it faster or parallelize it… but it seems to be possible.

### Whisper chunking

> **[Reason for 30s audio length · openai whisper · Discussion #1118](https://github.com/openai/whisper/discussions/1118)**
>
> Hello, I understand that whisper can only access 30s of audio content, what is the reason behind that? Is it because larger than 30s is harder to train? I assume a window larger than 30 seconds sin...

> **[Whisper chunking · openai whisper · Discussion #1977](https://github.com/openai/whisper/discussions/1977)**
>
> How does Whisper avoid dividing a word from the middle when it chunks audio into 30-seconds segments?

### Whisper optimizing

> **[Done! Accelerate Whisper tasks such as transcription, by multiprocesing...](https://github.com/openai/whisper/discussions/432)**
>
> I recently made my first ever commit, that is also one of my first Python programs, which I hope will be of benefit to others. The program accelerates Whisper tasks such as transcription, by multip...

> **[multiple GPUs · openai whisper · Discussion #360](https://github.com/openai/whisper/discussions/360)**
>
> Can multiple GPUs be used to load the model? FE: I have 2 GPUs (2x 8gb), with 1 I can load the medium model, but the large model crashes due to lack of memory. The model uses only 1 GPU, so is it p...

> **[speed up whisper? · openai whisper · Discussion #716](https://github.com/openai/whisper/discussions/716)**
>
> what are the best ways to squeeze as much performance out of whisper? we have been testingg it on various size GPU/CPUs but do not see an big difference from one to the other. Any ideas or pointers...