Dataset · danielrosehill
Read-aloud sentence recordings for ASR personalization
danielrosehill/Read-Aloud-Voice-Sentences
Single-speaker audio recordings in English and Hebrew with text transcripts for speech recognition fine-tuning.
What's inside
The dataset contains 10,000 to 100,000 audio-text pairs in English and Hebrew. It is a master set of read-aloud sentence recordings by a single speaker, collected via a dedicated voice capture app. The set serves as the primary source for smaller subsets and is designed for speech recognition system adaptation.
What you can do with it
- Personalization and fine-tuning of ASR models for a specific speaker's voice
- Training speech recognition in English and Hebrew
- Tasks
- Automatic speech recognition
- Languages
- English, Hebrew
- Size
- 10K–100K rows
- Downloads this month
- 60.3K
- Likes
- 0
- Access
- on request on Hugging Face
Models trained on it
- Fritter-ASR/FritterAutomatic speech recognition 1.6K
- mradermacher/Fritter-TTS-GGUF 549
- Hebrish/SabichAutomatic speech recognition 538
- Hebrish/JachnunAutomatic speech recognition 487
- Hebrish/ShakshukaAutomatic speech recognition 116
- Hebrish/Nemotron-Speech-Streaming-EN-0.6B-English-Hebrew-CodeswitchingAutomatic speech recognition 70
- Hebrish/Parakeet-TDT-0.6B-v2-Tech-And-AccentAutomatic speech recognition 57
- Fritter-ASR/Fritter-TTSText → speech 50
How to load it
from datasets import load_dataset
ds = load_dataset("danielrosehill/Read-Aloud-Voice-Sentences")