Dataset · reach-vb
30-hour female voice dataset for text-to-speech
reach-vb/jenny_tts_dataset
Various English texts read with an Irish accent for training TTS models.
What's inside
Contains approximately 30 hours of audio recordings and corresponding texts (10k to 100k records). The Irish-accented speaker Jenny read news headlines, excerpts from "1984" and "Little Women", Wikipedia articles, recipes, Reddit comments, song lyrics (including rap), and dialogues from the TV show "Friends". Audio is high quality: 48 kHz, 16-bit PCM.
What you can do with it
- Training and fine-tuning text-to-speech (TTS) models from scratch
- Creating a voice LoRA with natural intonation and an Irish accent
- Testing TTS on texts of various styles from news to rap
- Size
- 10K–100K rows
- Downloads this month
- 372
- Likes
- 38
Models trained on it
- 11mlabs/indri-0.1-124m-ttsText → speech 982
- mradermacher/VyvoTTS-LFM2-350M-Jenny-i1-GGUF 902
- parler-tts/parler-tts-mini-expressoText → speech 723
- mradermacher/VyvoTTS-LFM2-350M-Jenny-GGUF 306
- mradermacher/indri-0.1-124m-tts-GGUF 300
- mradermacher/indri-0.1-350m-tts-GGUF 262
- shangeth/Wren-TTS-0.5B-multi-expressiveText → speech 155
- parler-tts/parler-mini-v1-jennyText → speech 59
How to load it
from datasets import load_dataset
ds = load_dataset("reach-vb/jenny_tts_dataset")