Searcharxiv⌕ Search
Home/Archive/arXiv

arXiv · 2203.09825

AdaVocoder: Adaptive Vocoder for Custom Voice

Xin Yuan·Yongbing Feng·Mingming Ye·Cheng Tuo·Minghang Zhang
View original source ↗Read source document ↗Export metadata ↓

Abstract

Custom voice is to construct a personal speech synthesis system by adapting the source speech synthesis model to the target model through the target few recordings. The solution to constructing a custom voice is to combine an adaptive acoustic model with a robust vocoder. However, training a robust vocoder usually requires a multi-speaker dataset, which should include various age groups and various timbres, so that the trained vocoder can be used for unseen speakers. Collecting such a multi-speaker dataset is difficult, and the dataset distribution always has a mismatch with the distribution of the target speaker dataset. This paper proposes an adaptive vocoder for custom voice from another novel perspective to solve the above problems. The adaptive vocoder mainly uses a cross-domain consistency loss to solve the overfitting problem encountered by the GAN-based neural vocoder in the transfer learning of few-shot scenes. We construct two adaptive vocoders, AdaMelGAN and AdaHiFi-GAN. First, We pre-train the source vocoder model on AISHELL3 and CSMSC datasets, respectively. Then, fine-tune it on the internal dataset VXI-children with few adaptation data. The empirical results show that a high-quality custom voice system can be built by combining a adaptive acoustic model with a adaptive vocoder.

Explore related subjects

cs.SDcs.CLeess.AS

Keep this discovery

BibTeXRIS

Xin Yuan, Yongbing Feng, Mingming Ye, Cheng Tuo, Minghang Zhang. 2022-03-18. AdaVocoder: Adaptive Vocoder for Custom Voice. https://doi.org/10.21437/interspeech.2022-288

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

arXiv2026

Xiaomi-CocktailASR-1 Technical Report

Recently, large language model (LLM) based ASR models have achieved significant progress, yet they generally lack support for multi-speaker scenarios, where the cocktail party problem remains a critical bottleneck for further advancing ASR. Existing TS-ASR methods, including end-to-end architectures with speaker embeddings and latest LLM-based explorations suffer from degraded single-speaker performance and the inability to reject when the target speaker is absent. In this paper, we propose Xiaomi-CocktailASR-1, an LLM-based end-to-end TS-ASR architecture. By utilizing reference speech as voiceprint prompts, it directly transcribes the target speaker's speech without requiring speech separation. Xiaomi-CocktailASR-1 maintains competitive performance in single-speaker scenarios, comparable to mainstream ASR models. It also features a negative sample rejection capability, outputting empty text when the target speaker is absent from the mixed speech. Additionally, Xiaomi-CocktailASR-1 supports a Chain-of-Thought (CoT) reasoning mode to provide explicit reasoning steps. Extensive experiments on various synthetic and real-world multispeaker benchmarks demonstrate that Xiaomi-CocktailASR-1 achieves state-of-the-art performance, effectively addressing the cocktail party problem through a unified architecture that balances multispeaker and single-speaker recognition accuracy, along with rejection capability.

cs.SD↗
arXiv2026

EConv-TasNet: Efficient Conv-TasNet for Effective Speech Separation

Conv-TasNet has served as a strong baseline for time-domain speech separation, and many studies have extended it with advanced architectures such as dual-path networks, U-Nets, and attention mechanisms. However, these methods often introduce high computational cost and complexity, limiting their deployment in resource-constrained scenarios. To address this issue, we propose eConv-TasNet, an efficient variant of Conv-TasNet that improves both effectiveness and efficiency without relying on resource-intensive modules. The proposed model consists of a group-wise early-splitting (GES) module and a multi-group feature aggregation (MGFA) module. GES generates discriminative speaker embeddings at intermediate stages, while MGFA progressively aggregates these group-level representations for refined mask estimation. Experimental results show that eConv-TasNet reduces model size by 22.4%, accelerates inference by 18.9%, and improves SI-SNRi by 14.0%-28.0% across three public benchmarks. Moreover, it achieves competitive performance compared with state-of-the-art methods while requiring significantly fewer parameters and lower inference cost. These results demonstrate a favorable efficiency-effectiveness trade-off for edge deployment.

cs.SD↗