Realistic Multi-Microphone Data Simulation for Distant Speech Recognition

Abstract

The availability of realistic simulated corpora is of key importance for the future progress of distant speech recognition technology. The reliability, flexibility and low computational cost of a data simulation process may ultimately allow researchers to train, tune and test different techniques in a variety of acoustic scenarios, avoiding the laborious effort of directly recording real data from the targeted environment.In the last decade, several simulated corpora have been released to the research community, including the data-sets distributed in the context of projects and international challenges, such as CHiME and REVERB. These efforts were extremely useful to derive baselines and common evaluation frameworks for comparison purposes. At the same time, in many cases they highlighted the need of a better coherence between real and simulated conditions.In this paper, we examine this issue and we describe our approach to the generation of realistic corpora in a domestic context. Experimental validation, conducted in a multimicrophone scenario, shows that a comparable performance trend can be observed with both real and simulated data across different recognition frameworks, acoustic models, as well as multi-microphone processing techniques.

The Pytorch-kaldi Speech Recognition Toolkit

“…To validate our model in a more challenging scenario, experiments were also conducted in distant-talking conditions with the DIRHA-English dataset 4 [36,37]. Training was based on the original WSJ-5k corpus (consisting of 7, 138 sentences uttered by 83 speakers) that was contaminated with a set of impulse responses measured in a domestic environment [37]. The test phase was carried out with the real part of the dataset, consisting of 409 WSJ sentences uttered in the aforementioned environment by six native American speakers.…”

Predicting Word Error Rate for Reverberant Speech

“…The assumption is that much like blind or non-intrusive acoustic parameter estimation can be used as a proxy for estimating ASR performance [16], a neural network model can be trained to extract features from reverberant speech that are correlated with WER. The proposed method assumes reverberant speech samples transcribed by an ASR engine and the corresponding WER per utterance calculated by (7). The same data split as described in Section 2 is used.…”

Analysis of DNN Speech Signal Enhancement for Robust Speaker Recognition

“…To evaluate the impact of room acoustics on the accuracy of speaker verification, a proper dataset of reverberant audio is needed. An alternative that fills a qualitative gap between unsatisfying simulation (despite the improvement of realism reported in Ravanelli et al, 2016) and costly and demanding real speaker recording, is retransmission. To our advantage, we can also use the fact that a known dataset can be retransmitted so that the performances are readily comparable with known benchmarks.…”