Speech Synthesis: HiFi-GAN Model Card

Model Overview

HiFi-GAN is a neural vocoder model for text-to-speech applications. It is intended as the second part of a two-stage speech synthesis pipeline, with a mel-spectrogram generator such as FastPitch as the first stage.

Model Architecture

HiFi-GAN is a neural vocoder based on a generative adversarial network framework. During training, the model uses a powerful discriminator consisting of small sub-discriminators, each one focusing on specific periodic parts of a raw waveform. The generator is very fast and has a small footprint, while producing high quality speech.

Training

Dataset

This model is trained on proprietary data sampled at 44100Hz, and can be used to generate a Spanish (US) voice. This model supports 1 female and 1 male voice. The female voice comes with neutral, calm, angry and sad emotions. The male voice comes with neutral, calm, happy, and angry emotions. Each emotion is accessed as a speaker. For example Female-Calm, Male-Happy, and so on.

How to Use this Model

HiFi-GAN is intended to be used as the second part of a two stage speech synthesis pipeline. HiFi-GAN takes a mel-spectrogram and returns audio.

The encryption key for this model is tlt_encode

Input

Mel-spectrogram of shape (batch x mel_channels x time)

Output

Audio of shape (batch x time)

Limitations

N/A

References

[1] HiFi-GAN: Generative Adversarial Networks for Efficient and High Fidelity Speech Synthesis

Suggested Reading

Refer to the Riva documentation for more information.

License

By downloading and using the models and resources packaged with Riva Conversational AI, you accept the terms of the Riva license.

Ethical AI

NVIDIA’s platforms and application frameworks enable developers to build a wide array of AI applications. Consider potential algorithmic bias when choosing or creating the models being deployed. Work with the model’s developer to ensure that it meets the requirements for the relevant industry and use case; that the necessary instruction and documentation are provided to understand error rates, confidence intervals, and results; and that the model is being used under the conditions and in the manner intended.