Papers of the day   All papers

MelNet: A Generative Model for Audio in the Frequency Domain

Comments

Jun 05 2019 Kyle Kastner

Interested in a powerful new audio model for conditional and unconditional music, single, and multi-speaker TTS on in-the-wild data? Check out MelNet: https://arxiv.org/abs/1906.01083 Blog: https://sjvasquez.github.io/blog/melnet/ More samples: https://audio-samples.github.io/ Really incredible results! https://t.co/ZhTRSvWRA7
5 replies, 443 likes


Jun 06 2019 👩‍💻 DynamicWebPaige

"MelNet is broadly applicable to a variety of audio generation tasks—capable of unconditional speech generation, music generation, and text-to-speech synthesis, entirely end-to-end." 🎶 Demos: https://sjvasquez.github.io/blog/melnet/ 📰 Paper: https://arxiv.org/pdf/1906.01083.pdf Such realistic voices! 😮✨ https://t.co/YkEB1Rn0s4
1 replies, 51 likes


Jun 05 2019 Kyle McDonald

MelNet looks really promising for unconditional audio generation https://arxiv.org/abs/1906.01083 https://audio-samples.github.io
1 replies, 23 likes


Jun 05 2019 Statistics Papers

MelNet: A Generative Model for Audio in the Frequency Domain. http://arxiv.org/abs/1906.01083
0 replies, 21 likes


Jun 05 2019 Nal Kalchbrenner

Awesome work on a multi-scale approach to audio generation via Mel-spectrograms!
0 replies, 21 likes


Jun 05 2019 Norman Casagrande

This is really cool. "MelNet: A Generative Model for Audio in the Frequency Domain". It manages to capture prosody/style and long term dependencies seriously well. Kudos to the authors! paper: https://arxiv.org/pdf/1906.01083.pdf samples: https://audio-samples.github.io/
0 replies, 14 likes


Jun 05 2019 Stefan Lattner

Who would have thought - audio generation in the spectral domain using RNNs - amazing results!
0 replies, 2 likes


Jun 06 2019 최형석 (Hyeong-Seok Choi)

A big step on audio generative model
0 replies, 1 likes


Jun 06 2019 Grant Totten

Wow,the current state of the art in generative audio and speech synthesis.
0 replies, 1 likes


Jun 05 2019 sf

the WaveNet Baseline examples are amazing
0 replies, 1 likes


Jun 05 2019 Ozan Caglayan

If we will be able to prime a given TTS model very easily with in-the-wild speech samples, the implications of that would be immense. Such forged recordings can easily be used in Turkey for imprisonment.
1 replies, 0 likes


Content