A Comparison of Recent Neural Vocoders for Speech Signal Reconstruction (original) (raw)

High-quality Speech Synthesis Using Super-resolution Mel-Spectrogram

ArXiv, 2019

In speech synthesis and speech enhancement systems, melspectrograms need to be precise in acoustic representations. However, the generated spectrograms are over-smooth, that could not produce high quality synthesized speech. Inspired by image-to-image translation, we address this problem by using a learning-based post filter combining Pix2PixHD and ResUnet to reconstruct the mel-spectrograms together with super-resolution. From the resulting super-resolution spectrogram networks, we can generate enhanced spectrograms to produce high quality synthesized speech. Our proposed model achieves improved mean opinion scores (MOS) of 3.71 and 4.01 over baseline results of 3.29 and 3.84, while using vocoder Griffin-Lim and WaveNet, respectively.

Continuous vocoder in feed-forward deep neural network based speech synthesis

2018

Recently in statistical parametric speech synthesis, we proposed a vocoder using continuous F0 in combination with Maximum Voiced Frequency (MVF), which was successfully used with hidden Markov model (HMM) based text-to-speech (TTS). However, HMMs often generate over-smoothed and muffled synthesized speech. From this point, we propose here to use the modified version of our continuous vocoder with deep neural networks (DNNs) for further improving its quality. Evaluations between DNNTTS using Continuous and WORLD vocoders are also presented. Experimental results from objective and subjective tests have shown that the DNN-TTS have higher naturalness than HMM-TTS, and the proposed framework provides quality similar to the WORLD vocoder, while being simpler in terms of the number of excitation parameters and models better the voiced/unvoiced speech regions than the WORLD vocoder.

FeatherWave: An Efficient High-Fidelity Neural Vocoder with Multi-Band Linear Prediction

Interspeech 2020, 2020

In this paper, we propose the FeatherWave, yet another variant of WaveRNN vocoder combining the multi-band signal processing and the linear predictive coding. The LPCNet, a recently proposed neural vocoder which utilized the linear predictive characteristic of speech signal in the WaveRNN architecture, can generate high quality speech with a speed faster than real-time on a single CPU core. However, LPCNet is still not efficient enough for online speech generation tasks. To address this issue, we adopt the multi-band linear predictive coding for WaveRNN vocoder. The multi-band method enables the model to generate several speech samples in parallel at one step. Therefore, it can significantly improve the efficiency of speech synthesis. The proposed model with 4 sub-bands needs less than 1.6 GFLOPS for speech generation. In our experiments, it can generate 24 kHz high-fidelity audio 9x faster than realtime on a single CPU, which is much faster than the LPCNet vocoder. Furthermore, our subjective listening test shows that the FeatherWave can generate speech with better quality than LPCNet.

Parallel WaveNet: Fast High-Fidelity Speech Synthesis

Cornell University - arXiv, 2017

The recently-developed WaveNet architecture [27] is the current state of the art in realistic speech synthesis, consistently rated as more natural sounding for many different languages than any previous system. However, because WaveNet relies on sequential generation of one audio sample at a time, it is poorly suited to today's massively parallel computers, and therefore hard to deploy in a real-time production setting. This paper introduces Probability Density Distillation, a new method for training a parallel feed-forward network from a trained WaveNet with no significant difference in quality. The resulting system is capable of generating high-fidelity speech samples at more than 20 times faster than real-time, and is deployed online by Google Assistant, including serving multiple English and Japanese voices.

Universal Neural Vocoding with Parallel Wavenet

ICASSP 2021 - 2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

We present a universal neural vocoder based on Parallel WaveNet, with an additional conditioning network called Audio Encoder. Our universal vocoder offers real-time highquality speech synthesis on a wide range of use cases. We tested it on 43 internal speakers of diverse age and gender, speaking 20 languages in 17 unique styles, of which 7 voices and 5 styles were not exposed during training. We show that the proposed universal vocoder significantly outperforms speaker-dependent vocoders overall. We also show that the proposed vocoder outperforms several existing neural vocoder architectures in terms of naturalness and universality. These findings are consistent when we further test on more than 300 open-source voices.

On the Suitability of the Riesz Spectro-Temporal Envelope for WaveNet Based Speech Synthesis

Interspeech 2019

We address the problem of estimating the time-varying spectral envelope of a speech signal using a spectro-temporal demodulation technique. Unlike the conventional spectrogram, we consider a pitch-adaptive spectrogram and model a spectrotemporal patch using an amplitude-and frequency-modulated two-dimensional (2-D) cosine signal. We employ a demodulation technique based on the Riesz transform that we proposed recently to estimate the amplitude and frequency modulations. The amplitude modulation (AM) corresponds to the vocal-tract filter magnitude response (or envelope) and the frequency modulation (FM) corresponds to the excitation. We consider the AM and demonstrate its effectiveness by incorporating it as an acoustic feature for local conditioning in the statistical WaveNet vocoder for the task of speech synthesis. The quality of the synthesized speech obtained with the Riesz envelope is compared with that obtained using the envelope estimated by the WORLD vocoder. Objective measures and subjective listening tests on the CMU-Arctic database show that the quality of synthesis is superior to that obtained using the WORLD envelope. This study thus establishes the Riesz envelope as an efficient alternative to the WORLD envelope.

Towards Achieving Robust Universal Neural Vocoding

Interspeech 2019

This paper explores the potential universality of neural vocoders. We train a WaveRNN-based vocoder on 74 speakers coming from 17 languages. This vocoder is shown to be capable of generating speech of consistently good quality (98% relative mean MUSHRA when compared to natural speech) regardless of whether the input spectrogram comes from a speaker or style seen during training or from an out-of-domain scenario when the recording conditions are studio-quality. When the recordings show significant changes in quality, or when moving towards non-speech vocalizations or singing, the vocoder still significantly outperforms speaker-dependent vocoders, but operates at a lower average relative MUSHRA of 75%. These results are shown to be consistent across languages, regardless of them being seen during training (e.g. English or Japanese) or unseen (e.g. Wolof, Swahili, Ahmaric).

Efficient Neural Audio Synthesis

Cornell University - arXiv, 2018

Sequential models achieve state-of-the-art results in audio, visual and textual domains with respect to both estimating the data distribution and generating high-quality samples. Efficient sampling for this class of models has however remained an elusive problem. With a focus on text-to-speech synthesis, we describe a set of general techniques for reducing sampling time while maintaining high output quality. We first describe a single-layer recurrent neural network, the WaveRNN, with a dual softmax layer that matches the quality of the state-of-the-art WaveNet model. The compact form of the network makes it possible to generate 24 kHz 16-bit audio 4× faster than real time on a GPU. Second, we apply a weight pruning technique to reduce the number of weights in the WaveRNN. We find that, for a constant number of parameters, large sparse networks perform better than small dense networks and this relationship holds for sparsity levels beyond 96%. The small number of weights in a Sparse WaveRNN makes it possible to sample high-fidelity audio on a mobile CPU in real time. Finally, we propose a new generation scheme based on subscaling that folds a long sequence into a batch of shorter sequences and allows one to generate multiple samples at once. The Subscale WaveRNN produces 16 samples per step without loss of quality and offers an orthogonal method for increasing sampling efficiency. * Equal contribution 1 DeepMind 2 Google Brain.

WaveGrad 2: Iterative Refinement for Text-to-Speech Synthesis

Interspeech 2021, 2021

This paper introduces WaveGrad 2, a non-autoregressive generative model for text-to-speech synthesis. WaveGrad 2 is trained to estimate the gradient of the log conditional density of the waveform given a phoneme sequence. The model takes an input phoneme sequence, and through an iterative refinement process, generates an audio waveform. This contrasts to the original WaveGrad vocoder which conditions on mel-spectrogram features, generated by a separate model. The iterative refinement process starts from Gaussian noise, and through a series of refinement steps (e.g., 50 steps), progressively recovers the audio sequence. WaveGrad 2 offers a natural way to trade-off between inference speed and sample quality, through adjusting the number of refinement steps. Experiments show that the model can generate high fidelity audio, approaching the performance of a state-of-the-art neural TTS system. We also report various ablation studies over different model configurations. Audio samples are available at https://wavegrad.github.io/v2.

WaveNet-Based Speech Synthesis Applied to Czech

Text, Speech, and Dialogue, 2018

WaveNet is a recently-developed deep neural network for generating high-quality synthetic speech. It produces directly raw audio samples. This paper describes the first application of WaveNet-based speech synthesis for the Czech language. We used the basic WaveNet architecture. The duration of particular phones and the required fundamental frequency used for local conditioning were estimated by additional LSTM networks. We conducted a MUSHRA listening test to compare WaveNet with 2 traditional synthesis methods: unit selection and HMM-based synthesis. Experiments were performed on 4 large speech corpora. Though our implementation of WaveNet did not outperform the unit selection method as reported in other studies, there is still a lot of scope for improvement, while the unit selection TTS have probably reached its quality limit.