Restructuring speech representations using a pitch-adaptive time-frequency smoothing and an instantaneous-frequency-based F0 extraction: Possible role of a repetitive structure in sounds

By Hideki Kawahara, Ikuyo Masuda-Katsuse, A. de Cheveigné. 1999

Subfields: Natural Language Processing, Speech & Audio

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Natural Language Processing

1. Junichi Yamagishi
2. Tao Qin
3. Tie-Yan Liu
4. Xu Tan
5. Zhenhua Ling
6. Graham Neubig
7. Tomoki Toda
8. Yoshua Bengio

In Speech & Audio

1. Junichi Yamagishi
2. Xu Tan
3. Tomoki Toda
4. Tao Qin
5. Frank K. Soong
6. Tie-Yan Liu
7. Sheng Zhao
8. Lei He

Citations in the corpus

• WaveNet: A Generative Model for Raw Audio (2016, background): “There have been a number of attempts to address these issues individually, such as developing high-quality vocoders (Kawahara et al., 1999; Agiomyrgiannakis, 2015; Morise et al., 2016), improving the accuracy of generative models (Zen et al., 2007; 2013; Fan et al., 2014; Uria et al., 2015), and…”
• Vocos: Closing the gap between time-domain and Fourier-based neural vocoders for high-quality audio synthesis (2023, background): “While vocoders evolved significantly (Kawahara et al., 1999; Morise et al., 2016), they tended to oversimplify speech production, generating a distinctive ”buzzy” sound and thus compromising the naturalness of the speech.”
• A Survey on Neural Speech Synthesis (2021, methodology): “In vocoder analysis, it analyzes the speech and gets acoustic features such as mel-cepstral coefficients [79], band aperiodicity [153, 154] and F0.”
• Deep Voice 3: 2000-Speaker Neural Text-to-Speech (2017, background): “It is also worth noting that there are numerous alternatives for high-quality hand-engineered vocoders in the literature, such as STRAIGHT (Kawahara et al., 1999), Vocaine (Agiomyrgiannakis, 2015), and WORLD (Morise et al., 2016).”
• UnivNet: A Neural Vocoder with Multi-Resolution Spectrogram Discriminators for High-Fidelity Waveform Generation (2021, background): “Neural vocoders based on deep neural networks can generate human-like voices, instead of using traditional methods that contain audible artifacts[7, 8, 9].”
• Speaker Conditional WaveRNN: Towards Universal Neural Vocoder for Unseen Speaker and Recording Conditions (2020, methodology): “Traditional vocoder approaches mostly involved sourcefilter model for the generation of speech parameters [9, 10, 11, 12].”
• Text-to-Speech Synthesis (2009, methodology): “The STRAIGHT technique [189] is used for signal analysis and syn thesis (see Section 14.”
• v.morish'09: A Morphing-Based Singing Design Interface for Vocal Melodies (2009, methodology): “A high-quality vocoder, STRAIGHT [5], is used for the singing voice analysis, morphing and synthesis.”
• Statistical Parametric Speech Synthesis (2007, background): “Figure 11 shows a transitional segment of natural speech and three types of synthesized speech obtained by natural spectra and with the simple periodic pulse-train or white-noise excitation, the STRAIGHT’s excitation, and Maia et al.’s ML excitation modeling methods.”
• CheapTrick, a spectral envelope estimator for high-quality speech synthesis (2015, methodology): “A modern framework STRAIGHT (Kawahara et al., 1999) was proposed as an effective framework for highquality speech synthesis.”
• Lightweight LPCNet-Based Neural Vocoder with Tensor Decomposition (2020, methodology): “For comparison with a conventional signal-processing based vocoder, 40-dimensional mel-cepstrum, logarithmic F0, voiced unvoiced binary flag, and five-dimensional aperiodicity with 5 ms frame shift were extracted by STRAIGHT [3].”
• The HMM-based speech synthesis system (HTS) version 2.0 (2007, methodology): “1, with the STRAIGHT analysis /synthesis technique [82], will provide the ability to construct the state-ofthe-art HMM-based speech synthesis systems developed for the past Blizzard Challenge events [7,83].”
• Temporally variable multi-aspect N-way morphing based on interference-free speech representations (2013, background): “In this new formulation, derivative of mapping is log-linearly morphed, instead of morphing values directly.”
• An Investigation of Subband Wavenet Vocoder Covering Entire Audible Frequency Range with Limited Acoustic Features (2018, methodology): “To improve the synthesized speech quality in conventional SPSS and VC, sophisticated vocoders [3,4] have been introduced instead of a basic mel-log spectrum approximate (MLSA) filter with a simple pulse excitation and cepstrum [5].”
• Universal MelGAN: A Robust Neural Vocoder for High-Fidelity Waveform Generation in Multiple Domains (2020, unknown): “Neural vocoders generate humanlike voices using neural networks, instead of using traditional methods that contain audible artifacts [7, 8, 9].”