datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
indicf5-nfe10-100sentence-benchmark
Telugu TTS Benchmark (NFE=10, FP32)
This dataset contains 100 synthesized Telugu audio samples generated using an Indic TTS model with NFE=10 and FP32 precision on an RTX 5090.
Dataset Structure
The dataset uses the audiofolder format. The metadata.jsonl file located in the audio/ directory maps each .wav file to its corresponding Telugu text transcription, category, and generation metrics (RTF, VRAM, RMS).
Benchmark Configuration
NFE Step: 10… See the full description on the dataset page: https://huggingface.co/datasets/Noothi/indicf5-nfe10-100sentence-benchmark.telugu-indicf5-evaluationindicf5-nfe-9-10-11-stress-test