# Create speech to text training dataset using text to speech model

**URL:** <https://discuss.huggingface.co/t/create-speech-to-text-training-dataset-using-text-to-speech-model/31333>\
**Category:** Intermediate\
**Created:** [February 8, 2023, 3:44pm UTC](https://discuss.huggingface.co/t/create-speech-to-text-training-dataset-using-text-to-speech-model/31333 "2023-02-08T15:44:22Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![bjubert](https://avatars.discourse-cdn.com/v4/letter/b/a587f6/32.png) [@bjubert](https://discuss.huggingface.co/u/bjubert)\
**Post date:** [February 8, 2023, 3:44pm UTC](https://discuss.huggingface.co/t/create-speech-to-text-training-dataset-using-text-to-speech-model/31333/1 "2023-02-08T15:44:22Z")

</div>

Hello,

I want to use Whisper, a model for speech recognition but I have an issue. The problem is the way of people speak in my audio files is very specific so whisper encounters difficulties.

I had the idea to try to fine tune the model for better results but i don’t have enough data to do it. So my question is : is it possible to use a text to speech model to create a dataset to train a speech to text model ?

Thanks in advance.