EXPRESSO: A Benchmark and Analysis of Discrete Expressive Speech Resynthesis

By Tu Nguyen, Wei-Ning Hsu, Antony D'Avirro, Bowen Shi, Itai Gat, Maryam Fazel-Zarani, Tal Remez, Jade Copet, Gabriel Synnaeve, Michael Hassid and 3 more. 2023

130 citations recorded by the source corpus.

Recent work has shown that it is possible to resynthesize high-quality speech based, not on text, but on low bitrate discrete units that have been learned in a self-supervised fashion and can therefore capture expressive aspects of speech that are hard to transcribe (prosody, voice styles, non-verbal vocalization). The adoption of these methods is still limited by the fact that most speech synthesis datasets are read, severely limiting spontaneity and expressivity. Here, we introduce Expresso, a high-quality expressive speech dataset for textless speech synthesis that includes both read speech and improvised dialogues rendered in 26 spontaneous expressive styles. We illustrate the challenges and potentials of this dataset with an expressive resynthesis benchmark where the task is to encode the input in low-bitrate units and resynthesize it in a target voice while preserving content and style. We evaluate resynthesis quality with automatic metrics for different self-supervised discrete encoders, and explore tradeoffs between quality, bitrate and invariance to speaker and style. All the dataset, evaluation metrics and baseline models are open source

Subfields: Artificial Intelligence, Experimental and Cognitive Psychology

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Chenpeng Du
2. Xie Chen
3. K. Yu
4. Andros Tjandra
5. Apoorv Vyas
6. Daniel Povey
7. Ziyang Ma
8. Yifan Yang

In Experimental and Cognitive Psychology

1. Chenpeng Du
2. Xie Chen
3. K. Yu

Citations in the corpus

• Audiobox: Unified Audio Generation with Natural Language Prompts (2023, background): “This feature is often desired as exemplified in earlier work where emotion and voice can be controlled independently (Hsu et al., 2019; Kulkarni et al., 2021; Nguyen et al., 2023).”
• Towards Universal Speech Discrete Tokens: A Case Study for ASR and TTS (2023, methodology): “Nguyen et al. [16] utilized discrete tokens from Hu-BERT and Encodec for speech resynthesis and found Encodec, designed for general audio compression, outperformed HuBERT units in this task.”
• HierSpeech++: Bridging the Gap Between Semantic and Acoustic Representation of Speech by Hierarchical Variational Inference for Zero-Shot Speech Synthesis (2023, unknown): “We used a whisper and conversational speech of EXPRESSO [77] by splitting the segment using VAD information for each speaker and NIKL.”