# Problem-Agnostic Speech Embeddings for Multi-Speaker Text-to-Speech with SampleRNN

arXiv: https://arxiv.org/abs/1906.00733 · HTML: https://mlsanity.com/paper/1906.00733/

## Key facts

- Cited by 6 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 3 Jun 2019.

## Built on by (most cited first, 6 of 6)

- [SUPERB: Speech processing Universal PERformance Benchmark](https://mlsanity.com/paper/2105.01051/) (2021)
- [SUPERB-SG: Enhanced Speech processing Universal PERformance Benchmark for Semantic and Generative Capabilities](https://mlsanity.com/paper/2203.06849/) (2022)
- [Speech Self-Supervised Representation Benchmarking: Are We Doing it Right?](https://mlsanity.com/paper/2306.00452/) (2023)
- [Self-Supervised Representation Learning for Speech Using Visual Grounding and Masked Language Modeling](https://mlsanity.com/paper/2202.03543/) (2022)
- [Improving Noise Robustness of Contrastive Speech Representation Learning with Speech Reconstruction](https://mlsanity.com/paper/2110.15430/) (2021)
- [On permutation invariant training for speech source separation](https://mlsanity.com/paper/2102.04945/) (2021)

## Abstract (arXiv)

Text-to-speech (TTS) acoustic models map linguistic features into an acoustic representation out of which an audible waveform is generated. The latest and most natural TTS systems build a direct mapping between linguistic and waveform domains, like SampleRNN. This way, possible signal naturalness losses are avoided as intermediate acoustic representations are discarded. Another important dimension of study apart from naturalness is their adaptability to generate voice from new speakers that were unseen during training. In this paper we first propose the use of problem-agnostic speech embeddings in a multi-speaker acoustic model for TTS based on SampleRNN. This way we feed the acoustic model with speaker acoustically dependent representations that enrich the waveform generation more than discrete embeddings unrelated to these factors. Our first results suggest that the proposed embeddings lead to better quality voices than those obtained with discrete embeddings. Furthermore, as we can use any speech segment as an encoded representation during inference, the model is capable to generalize to new speaker identities without retraining the network. We finally show that, with a small increase of speech duration in the embedding extractor, we dramatically reduce the spectral distortion to close the gap towards the target identities.