Small-E: Small Language Model with Linear Attention for Efficient Speech Synthesis - Institut de Recherche et Coordination Acoustique/Musique Accéder directement au contenu
Communication Dans Un Congrès Année : 2024

Small-E: Small Language Model with Linear Attention for Efficient Speech Synthesis

Théodor Lemerle
  • Fonction : Auteur
  • PersonId : 1392154
Nicolas Obin
Axel Roebel

Résumé

Recent advancements in text-to-speech (TTS) powered by language models have showcased remarkable capabilities in achieving naturalness and zero-shot voice cloning. Notably, the decoder-only transformer is the prominent architecture in this domain. However, transformers face challenges stemming from their quadratic complexity in sequence length, impeding training on lengthy sequences and resource-constrained hardware. Moreover they lack specific inductive bias with regards to the monotonic nature of TTS alignments. In response, we propose to replace transformers with emerging recurrent architectures and introduce specialized cross-attention mechanisms for reducing repeating and skipping issues. Consequently our architecture can be efficiently trained on long samples and achieve state-of-the-art zero-shot voice cloning against baselines of comparable size. Our implementation and demos are available at \url{https://github.com/theodorblackbird/lina-speech}.
Fichier principal
Vignette du fichier
Small_E.pdf (350.6 Ko) Télécharger le fichier
Origine Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-04611889 , version 1 (14-06-2024)

Identifiants

  • HAL Id : hal-04611889 , version 1

Citer

Théodor Lemerle, Nicolas Obin, Axel Roebel. Small-E: Small Language Model with Linear Attention for Efficient Speech Synthesis. Interspeech, ISCA, Sep 2024, Kos Island, Greece. ⟨hal-04611889⟩
0 Consultations
0 Téléchargements

Partager

Gmail Mastodon Facebook X LinkedIn More