Hugging Face's logo

Datasets:
Noothi
/
telugu-tech-indicf5-custom-voice-v2

Tasks:
Text-to-Speech
Automatic Speech Recognition
Languages:
Telugu
Size:
n<1K
Tags:
tts
telugu
indicf5
f5-tts
speech
elevenlabs
License:
Dataset card Files
xet
 Community
Dataset Viewer
The dataset could not be loaded because the splits use different data file formats, which is not supported. Read more about the splits configuration. Click for more details.
Couldn't infer the same data file format for all splits. Got {NamedSplit('train'): ('csv', {}), NamedSplit('validation'): ('audiofolder', {})}
Error code:   FileFormatMismatchBetweenSplitsError

Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.

🎙️ Telugu Technical Custom Voice Dataset

A high-quality, single-speaker Telugu tech speech dataset designed for fine-tuning text-to-speech (TTS) models like IndicF5-TTS, F5-TTS, XTTS v2, VITS, and ElevenLabs Voice Cloning.

📊 Dataset Overview

  • Total Clips: 676 WAV files
  • Total Audio Duration: 70.61 minutes (1.18 hours / 4,236.54 seconds)
  • Total Disk Size: 1.14 GB
  • Average Clip Duration: 6.26 seconds (ranging 2.0s – 15.0s, optimal for TTS attention alignment)
  • Audio Format: 22,050 Hz / 48,000 Hz, 16-bit PCM Mono
  • STT Transcription Engine: ElevenLabs Scribe V2 (scribe_v2)
  • Train / Validation Split: 658 Train clips / 18 Validation clips

🎬 Video Source Breakdown

Video Title WAV Clips Duration Disk Size
Are we writing IF ELSE conditions properly ? 171 clips 17.80 mins 293.52 MB
Production Backend Systems Part 1 163 clips 15.66 mins 258.29 MB
Excel in software dev without CS degree 149 clips 17.68 mins 291.58 MB
Why Netflix uses HTTP3 (Part 1) 122 clips 12.13 mins 200.06 MB
How 1 thread handles 1M reqs (Node.js, Redis) 72 clips 7.35 mins 121.18 MB
TOTAL DATASET 676 clips 70.61 mins 1.14 GB

📁 Files Structure

  • wavs/: Isolated voice clips in .wav format.
  • metadata.csv: Raw ElevenLabs Scribe V2 transcriptions (| delimited).
  • metadata_clean.csv: Cleaned, NFC-normalized text with Code-Switched technical terms.
  • metadata_telugu.csv: Pure Telugu script phonetic transliterations (, delimited).
  • train.csv: Training split (658 clips).
  • val.csv: Validation split (18 clips).

🚀 Usage with Hugging Face Datasets 

from datasets import load_dataset

dataset = load_dataset("Noothi/telugu-tech-custom-voice-v2")
print(dataset)
Downloads last month
47
Total file size:
1.22 GB

Models trained or fine-tuned on Noothi/telugu-tech-indicf5-custom-voice-v2