Multiband excitation vocoder

By Daniel W. Griffin, Jae S. Lim. 1988

A speech model, referred to as the multiband excitation model, is presented. In this model the band around each harmonic of the fundamental frequency is declared voiced or unvoiced. Estimation methods for the parameters of the model are developed and methods to synthesize speech from the model parameters are described. To illustrate a potential application of the speech model, an 8 kb/s vocoder is developed and its performance is evaluated. Both informal listening and intelligibility tests show that the vocoder has very good performance both in speech quality and intelligibility, particularly for noisy speech.<<ETX>>

How researchers see this paper

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

Citations in the corpus

Wavenet Based Low Rate Speech Coding (2017, background): “The voicing level is determined as in the multiband excitation vocoder [21].”
Neural Source-Filter Waveform Models for Statistical Parametric Speech Synthesis (2019, methodology): “Although spectral amplitude distance has been used in classical speech coding methods [30], [31], [32], we further compute multiple distances with different short-time analysis configurations, as illustrated in Figure 2.”
Composition of Deep and Spiking Neural Networks for Very Low Bit Rate Speech Coding (2016, methodology): “The phonetic encoder — automatic speech recognition (ASR) — consists of acoustic HMMs and language models and an incremental search module.”
Neural Source-filter-based Waveform Model for Statistical Parametric Speech Synthesis (2018, background): “It is also possible to try classical speech modeling methods, including glottal waveform excitations [22, 23], two-bands or multi-bands approaches [24, 25] on waveforms.”
Text-to-Speech Synthesis (2009, unknown): “A number of techniques based on this principle have been proposed, include the multi-band excitation (MBE) model [185], [281] and a parallel hybrid approach [2].”
Applying the harmonic plus noise model in concatenative speech synthesis (2001, methodology): “First, an initial pitch estimation is obtained by searching the minimum value of an error function, as proposed in [19], over a prespecified set of pitch periods.”
Techniques for Harmonic Sinusoidal Coding (1997, methodology): “The estimation of the harmonic magnitudes and phases depends on the accurate estimation of the fundamental, ω0 l [28].”
Pitch-synchronous waveform processing techniques for text-to-speech synthesis using diphones (1989, background): “…from cycle to cycle (which can be observed on real speech): these variations can be attributed partly to irregularities in the vocal cord movement and partly to the turbulent airflow from the lungs during the open-glottis period in each pitch cycle (Kang and Everett, 1985; Griffin and Lim, 1988).”
Vocaine the vocoder and applications in speech synthesis (2015, background): “the Sinusoidal Transform Codec (STC) [2, 3] and Advanced MultiBand Excitation (AMBE) [4] but was later reduced for about a decade following the dominance of CELP codecs.”
Robust Speaker-Adaptive HMM-Based Text-to-Speech Synthesis (2009, background): “To reduce buzziness, mixed or multi-band excitation techniques [15]–[17] have been integrated into the basic systems to replace the simple pulse or noise excitation and have been evaluated [18]–[21].”
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 (1999, methodology): “More sophisticated signal models like Multi Band Excitation (MBE) (Grin and Lim, 1988; Dutoit and Leich, 1993), multi-pulse (Caspers and Atal, 1987) and others (Abrantes et al.”