Skip to content

Dataset · HumynLabs

Audio recordings for ASR training

HumynLabs/speech-recognition-data

Short standard text recordings for training and evaluating speech recognition.

What's inside

The dataset consists of high-quality audio recordings reading a standard 50-word paragraph. The recordings are intended for training and evaluating ASR models. Volume is under 1K examples; the recording text is in English.

What you can do with it

  • Fine-tuning ASR models on clean speech
  • Evaluating recognition of a standard phrase
  • Collecting voice data for a voice assistant
Size
up to 1K rows
Downloads this month
315
Likes
478

How to load it

from datasets import load_dataset

ds = load_dataset("HumynLabs/speech-recognition-data")