vocoder
Datasets
All datasets matching “vocoder”rhea-tts-vocoder
Unified Vocoder Training Dataset
This dataset combines LJSpeech, VCTK, and LibriTTS for training neural vocoders. Each dataset is available as a separate configuration, with a default 'all' configuration that combines all datasets. You can train on individual datasets or use the combined dataset as needed.
Dataset Statistics
Total files: 43,691
Total duration: 71.2 hours
Total speakers: 326
Per-dataset breakdown:
libritts:
Files: 43,691
Duration: 71.2 hours… See the full description on the dataset page: https://huggingface.co/datasets/consulted-graphs/rhea-tts-vocoder.Kouon_NSF_Vocoder-training_datainterspeech2024_discrete_speech_vocoder_resultsNSF_Vocoder_datasetgenerated-audio-vocoder-MelGAN-HiFiGAN-PWG-MultibandMelGAN-FullbandMelGAN-WaveGlowThis dataset includes samples of the vocoder of MelGAN, HiFi-GAN, Parallel WaveGAN (PWG), Multi-band MelGAN, Full-band MelGAN, and WaveGlow.
interspeech2024_discrete_speech_vocoder_requests