Stepback trains a voice converter with two decoders and a self-destructive loss to separate speaker identity from linguistic content, but the preprint contains no reported evaluation results.
CycleGAN-VC2: Improved CycleGAN-based Non-parallel Voice Conversion
1 Pith paper cite this work. Polarity classification is still indexing.
abstract
Non-parallel voice conversion (VC) is a technique for learning the mapping from source to target speech without relying on parallel data. This is an important task, but it has been challenging due to the disadvantages of the training conditions. Recently, CycleGAN-VC has provided a breakthrough and performed comparably to a parallel VC method without relying on any extra data, modules, or time alignment procedures. However, there is still a large gap between the real target and converted speech, and bridging this gap remains a challenge. To reduce this gap, we propose CycleGAN-VC2, which is an improved version of CycleGAN-VC incorporating three new techniques: an improved objective (two-step adversarial losses), improved generator (2-1-2D CNN), and improved discriminator (PatchGAN). We evaluated our method on a non-parallel VC task and analyzed the effect of each technique in detail. An objective evaluation showed that these techniques help bring the converted feature sequence closer to the target in terms of both global and local structures, which we assess by using Mel-cepstral distortion and modulation spectra distance, respectively. A subjective evaluation showed that CycleGAN-VC2 outperforms CycleGAN-VC in terms of naturalness and similarity for every speaker pair, including intra-gender and inter-gender pairs.
citation-role summary
citation-polarity summary
fields
cs.SD 1years
2025 1verdicts
REJECT 1roles
background 1polarities
background 1representative citing papers
citing papers explorer
-
Stepback: Enhanced Disentanglement for Voice Conversion via Multi-Task Learning
Stepback trains a voice converter with two decoders and a self-destructive loss to separate speaker identity from linguistic content, but the preprint contains no reported evaluation results.