SPADE: Structured Pruning and Adaptive Distillation for Efficient LLM-TTS
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2025
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866912856747802624 |
|---|---|
| author | Nguyen, Tan Dat Kim, Jaehun Kim, Ji-Hoon Choi, Shukjae Lim, Youshin Chung, Joon Son |
| author_facet | Nguyen, Tan Dat Kim, Jaehun Kim, Ji-Hoon Choi, Shukjae Lim, Youshin Chung, Joon Son |
| contents | The goal of this paper is to introduce SPADE, a framework for Structured Pruning and Adaptive Distillation for Efficient Large Language Model-based text-to-speech (LLM-TTS). Recent LLM-TTS systems achieve strong controllability and zero-shot generalization, but their large parameter counts and high latency limit real-world deployment. SPADE addresses this by combining (i) a pruning step guided by a word-error-rate-based layer importance index to remove non-essential Transformer layers, with (ii) multi-level knowledge distillation to restore autoregressive coherence. On zero-shot benchmarks, SPADE preserves near-parity perceptual quality while halving Transformer depth, reducing VRAM usage by up to 20%, and achieving up to 1.7x faster real-time factor with less than 5% of the original training data. These results show that compact LLM-TTS models can maintain naturalness and speaker similarity while enabling practical real-time speech generation. Audio samples are available at https://mm.kaist.ac.kr/projects/SPADE/. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2509_20802 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | SPADE: Structured Pruning and Adaptive Distillation for Efficient LLM-TTS Nguyen, Tan Dat Kim, Jaehun Kim, Ji-Hoon Choi, Shukjae Lim, Youshin Chung, Joon Son Audio and Speech Processing Sound The goal of this paper is to introduce SPADE, a framework for Structured Pruning and Adaptive Distillation for Efficient Large Language Model-based text-to-speech (LLM-TTS). Recent LLM-TTS systems achieve strong controllability and zero-shot generalization, but their large parameter counts and high latency limit real-world deployment. SPADE addresses this by combining (i) a pruning step guided by a word-error-rate-based layer importance index to remove non-essential Transformer layers, with (ii) multi-level knowledge distillation to restore autoregressive coherence. On zero-shot benchmarks, SPADE preserves near-parity perceptual quality while halving Transformer depth, reducing VRAM usage by up to 20%, and achieving up to 1.7x faster real-time factor with less than 5% of the original training data. These results show that compact LLM-TTS models can maintain naturalness and speaker similarity while enabling practical real-time speech generation. Audio samples are available at https://mm.kaist.ac.kr/projects/SPADE/. |
| title | SPADE: Structured Pruning and Adaptive Distillation for Efficient LLM-TTS |
| topic | Audio and Speech Processing Sound |
| url | https://arxiv.org/abs/2509.20802 |