We have previously proposed a method that allows for non-parallel voice conversion (VC) by using a variant of generative adversarial networks (GANs) called StarGAN. The main features of our method, called StarGAN-VC, are as follows: First, it requires no parallel utterances, transcriptions, or time alignment procedures for speech generator training. Second, it can simultaneously learn mappings across multiple domains using a single generator network so that it can fully exploit available training data collected from multiple domains to capture latent features that are common to all the domains. Third, it is able to generate converted speech signals quickly enough to allow real-time implementations and requires only several minutes of training examples to generate reasonably realistic-sounding speech. In this paper, we describe three formulations of StarGAN, including a newly introduced novel StarGAN variant called "Augmented classifier StarGAN (A-StarGAN)", and compare them in a non-parallel VC task. We also compare them with several baseline methods.

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 (twostep 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. 1

No abstract

Eye and head morphology vary considerably among insects and even between closely related species of Drosophila. Species of the D. melanogaster subgroup, and other Drosophila species, exhibit a negative correlation between eye size and face width (FW); for example, D. mauritiana generally has bigger eyes composed of larger ommatidia and conversely a narrower face than its sibling species. To better understand the evolution of eye and head morphology, we investigated the genetic and developmental basis of differences in eye size and FW between male D. mauritiana and D. simulans. QTL mapping of eye size and FW showed that the major loci responsible for the interspecific variation in these traits are localized to different genomic regions. Introgression of the largest effect QTL underlying the difference in eye size resulted in flies with larger eyes but no significant difference in FW. Moreover, introgression of a QTL region on the third chromosome that contributes to the FW difference between these species affected FW, but not eye size. We also observed that this difference in FW is detectable earlier in the development of the eye-antennal disc than the difference in the size of the retinal field. Our results suggest that different loci that act at different developmental stages underlie changes in eye size and FW. Therefore, while there is a negative correlation between these traits in Drosophila, we show genetically that they also have the potential to evolve independently and this may help to explain the evolution of these traits in other insects.

This paper describes a method based on a sequenceto-sequence learning (Seq2Seq) with attention and context preservation mechanism for voice conversion (VC) tasks. Seq2Seq has been outstanding at numerous tasks involving sequence modeling such as speech synthesis and recognition, machine translation, and image captioning. In contrast to current VC techniques, our method 1) stabilizes and accelerates the training procedure by considering guided attention and proposed context preservation losses, 2) allows not only spectral envelopes but also fundamental frequency contours and durations of speech to be converted, 3) requires no context information such as phoneme labels, and 4) requires no time-aligned source and target speech data in advance. In our experiment, the proposed VC framework can be trained in only one day, using only one GPU of an NVIDIA Tesla K80, while the quality of the synthesized speech is higher than that of speech converted by Gaussian mixture model-based VC and is comparable to that of speech generated by recurrent neural network-based text-to-speech synthesis, which can be regarded as an upper limit on VC performance.

Male sexual characters are often among the first traits to diverge between closely related species and identifying the genetic basis of such changes can contribute to our understanding of their evolutionary history. However, little is known about the genetic architecture or the specific genes underlying the evolution of male genitalia. The morphology of the claspers, posterior lobes, and anal plates exhibit striking differences between Drosophila mauritiana and D. simulans. Using QTL and introgression-based high-resolution mapping, we identified several small regions on chromosome arms 3L and 3R that contribute to differences in these traits. However, we found that the loci underlying the evolution of clasper differences between these two species are independent from those that contribute to posterior lobe and anal plate divergence. Furthermore, while most of the loci affect each trait in the same direction and act additively, we also found evidence for epistasis between loci for clasper bristle number. In addition, we conducted an RNAi screen in D. melanogaster to investigate if positional and expression candidate genes located on chromosome 3L, are also involved in genital development. We found that six of these genes, including components of Wnt signaling and male-specific lethal 3 (msl3), regulate the development of genital traits consistent with the effects of the introgressed regions where they are located and that thus represent promising candidate genes for the evolution these traits.

This paper proposes a non-parallel many-to-many voice conversion (VC) method using a variant of the conditional variational autoencoder (VAE) called an auxiliary classifier VAE (ACVAE). The proposed method has three key features. First, it adopts fully convolutional architectures to construct the encoder and decoder networks so that the networks can learn conversion rules that capture time dependencies in the acoustic feature sequences of source and target speech. Second, it uses an information-theoretic regularization for the model training to ensure that the information in the attribute class label will not be lost in the conversion process. With regular CVAEs, the encoder and decoder are free to ignore the attribute class label input. This can be problematic since in such a situation, the attribute class label will have little effect on controlling the voice characteristics of input speech at test time. Such situations can be avoided by introducing an auxiliary classifier and training the encoder and decoder so that the attribute classes of the decoder outputs are correctly predicted by the classifier. Third, it avoids producing buzzy-sounding speech at test time by simply transplanting the spectral details of the input speech into its converted version. Subjective evaluation experiments revealed that this simple method worked reasonably well in a non-parallel many-to-many speaker identity conversion task.

Non-parallel multi-domain voice conversion (VC) is a technique for learning mappings among multiple domains without relying on parallel data. This is important but challenging owing to the requirement of learning multiple mappings and the nonavailability of explicit supervision. Recently, StarGAN-VC has garnered attention owing to its ability to solve this problem only using a single generator. However, there is still a gap between real and converted speech. To bridge this gap, we rethink conditional methods of StarGAN-VC, which are key components for achieving non-parallel multi-domain VC in a single model, and propose an improved variant called StarGAN-VC2. Particularly, we rethink conditional methods in two aspects: training objectives and network architectures. For the former, we propose a source-and-target conditional adversarial loss that allows all source domain data to be convertible to the target domain data. For the latter, we introduce a modulation-based conditional method that can transform the modulation of the acoustic feature in a domain-specific manner. We evaluated our methods on non-parallel multi-speaker VC. An objective evaluation demonstrates that our proposed methods improve speech quality in terms of both global and local structure measures. Furthermore, a subjective evaluation shows that StarGAN-VC2 outperforms StarGAN-VC in terms of naturalness and speaker similarity. 1