Speech Enhancement With Inventory Style Speech Resynthesis

Xiao, Xiong; Nickel, Robert M.

doi:10.1109/tasl.2009.2031793

Cited by 26 publications

(20 citation statements)

References 41 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This approach is inspired by several speech processing systems that use codebooks to constrain estimates to be within a subset of all possible values. For example, inventory-style enhancement maps input noisy mel-frequency cepstral coefficient (MFCC) vectors to clean MFCC vectors via a joint codebook trained on clean and noisy speech [19]. Codebooks have also been used to provide maximum-likelihood estimates of speech and noise linear predictor parameters by searching for the combination of codebook entries that maximise the likelihood [20].…”

Section: Introductionmentioning

confidence: 99%

Generating Intelligible Audio Speech From Visual Speech

Cornu

Milner

2017

IEEE/ACM Trans. Audio Speech Lang. Process.

View full text Add to dashboard Cite

Abstract-This work is concerned with generating intelligible audio speech from a video of a person talking. Regression and classification methods are proposed first to estimate static spectral envelope features from active appearance model (AAM) visual features. Two further methods are then developed to incorporate temporal information into the prediction -a feature-level method using multiple frames and a model-level method based on recurrent neural networks. Speech excitation information is not available from the visual signal, so methods to artificially generate aperiodicity and fundamental frequency are developed. These are combined within the STRAIGHT vocoder to produce a speech signal. The various systems are optimised through objective tests before applying subjective intelligibility tests that determine a word accuracy of 85% from a set of human listeners on the GRID audio-visual speech database. This compares favourably with a previous regression-based system that serves as a baseline which achieved a word accuracy of 33%.

show abstract

Section: Introductionmentioning

confidence: 99%

Generating Intelligible Audio Speech From Visual Speech

Cornu

Milner

2017

IEEE/ACM Trans. Audio Speech Lang. Process.

View full text Add to dashboard Cite

show abstract

“…Several methods have been proposed recently for corpusbased speech enhancement [7][8][9][10][11]. Work in one research thread [7][8][9] builds dictionaries of noisy speech, adding noise to known clean speech.…”

Section: Related Workmentioning

confidence: 99%

“…Work in one research thread [7][8][9] builds dictionaries of noisy speech, adding noise to known clean speech. While this makes comparing the observation to the dictionary simple, it requires a very large dictionary and access to the noise at training time.…”

Section: Related Workmentioning

confidence: 99%

Learning a concatenative resynthesis system for noise suppression

Mandel

Cho

Wang

2014

2014 IEEE Global Conference on Signal and Information Processing (GlobalSIP)

View full text Add to dashboard Cite

Abstract-This paper introduces a new approach to dictionarybased source separation employing a learned non-linear metric. In contrast to existing parametric source separation systems, this model is able to utilize a rich dictionary of speech signals. In contrast to previous dictionary-based source separation systems, the system can utilize perceptually relevant non-linear features of the noisy and clean audio. This approach utilizes a deep neural network (DNN) to predict whether a noisy chunk of audio contains a given clean chunk. Speaker-dependent experiments on the small-vocabulary CHiME2-GRID corpus show that this model is able to accurately resynthesize clean speech from noisy observations. Preliminary listening tests show that the system's output has much higher audio quality than existing parametric systems trained on the same data, achieving noise suppression levels close to those of the original clean speech.

show abstract

“…More recently, several approaches have been proposed that instead synthesise or reconstruct a clean speech signal. For example, corpus and inventory methods use noisy speech to identify segments from a database of clean speech which are concatenated to form the enhanced speech signal [6,7]. Methods that utilise HMMs within the enhancement process fall into both the filtering and reconstruction approaches.…”

Section: Introductionmentioning

confidence: 99%

HMM-Based Speech Enhancement Using Sub-Word Models and Noise Adaptation

Kato¹,

Milner²

2016

Interspeech 2016

View full text Add to dashboard Cite

This work proposes a method of speech enhancement that uses a network of HMMs to first decode noisy speech and to then synthesise a set of features that enables a clean speech signal to be reconstructed. Different choices of acoustic model (whole-word, monophone and triphone) and grammars (highly constrained to no constraints) are considered and the effects of introducing or relaxing acoustic and grammar constraints investigated. For robust operation in noisy conditions it is necessary for the HMMs to model noisy speech and consequently noise adaptation is investigated along with its effect on the reconstructed speech. Speech quality and intelligibility analysis find triphone models with no grammar, combined with noise adaptation, gives highest performance that outperforms conventional methods of enhancement at low signal-to-noise ratios.

show abstract

Speech Enhancement With Inventory Style Speech Resynthesis

Cited by 26 publications

References 41 publications

Generating Intelligible Audio Speech From Visual Speech

Generating Intelligible Audio Speech From Visual Speech

Learning a concatenative resynthesis system for noise suppression

HMM-Based Speech Enhancement Using Sub-Word Models and Noise Adaptation

Contact Info

Product

Resources

About