The reverb challenge: A common evaluation framework for dereverberation and recognition of reverberant speech

Abstract

Recently, substantial progress has been made in the field of reverberant speech signal processing, including both single-and multichannel de-reverberation techniques, and automatic speech recognition (ASR) techniques robust to reverberation. To evaluate state-ofthe-art algorithms and obtain new insights regarding potential future research directions, we propose a common evaluation framework including datasets, tasks, and evaluation metrics for both speech enhancement and ASR techniques. The proposed framework will be used as a common basis for the REVERB (REverberant Voice Enhancement and Recognition Benchmark) challenge. This paper describes the rationale behind the challenge, and provides a detailed description of the evaluation framework and benchmark results.

An analysis of environment, microphone and data simulation mismatches in robust speech recognition

“…The adaptation of DNN acoustic models to specific acoustic conditions has been investigated, e.g., (Seltzer et al, 2013;Karanasou et al, 2014), however it has been evaluated in multi-condition settings rather than actual mismatched conditions. The impact of the number of microphones on the WER obtained after enhancing reverberated speech was evaluated in the REVERB challenge (Kinoshita et al, 2013), but the impact of microphone distance was not considered and no such large-scale experiment was performed with noisy speech. To our knowledge, a study of the impact of mismatched noise environments on the resulting ASR performance is also missing.…”

“…Few existing datasets involve both real and simulated data. In the REVERB dataset (Kinoshita et al, 2013), the speaker distances for real and simulated data differ, which does not allow fair comparison. The CHiME-3 dataset (Barker et al, 2015) provides a data simulation tool which aims to reproduce the characteristics of real data for training and twinned real and simulated data pairs for development and testing.…”

Probabilistic integration of diffuse noise suppression and dereverberation

“…We took observed signals from the REVERB challenge database [19], specifically the real-world eight-channel recording AMI WSJ20-Array1-* T10c020c.wav in RealData. The recording contains reverberation of RT60 ∼ 0.7 s and some noise, and was truncated at 2 s. The sampling frequency was 16 kHz.…”

“…The frame length and hop for STFT were 1024 and 256 points (equivalent to 64 and 16 ms), respectively; the window type Hamming; the prediction order K = 3; the prediction delay Δ = 3; the number of iterations 20. We evaluate the following measures as defined in [19]: the cepstrum distance (CD), the log-likelihood ratio (LLR), the frequency-weighted segmental signal-to-noise ratio (FWSegSNR), and the speech-to-reverberation modulation energy ratio (SRMR). These measures were evaluated by using the headset recording AMI WSJ20-Headset1 T10c020c.wav as reference.…”

Prior-based Binary Masking and Discriminative Methods for Reverberant and Noisy Speech Recognition Using Distant Stereo Microphones

“…In such scenarios, close-talking input is often impractical or unsafe, and, while very challenging, allowing the speaker to be far from the microphone is highly desirable. To validate the effectiveness of state-of-the-art speech enhancement and ASR techniques in distant-talking conditions, several challenges have been organized [4], [22], [44]. Among these, the Computational Hearing in Multisource Environments (CHiME) challenges recently introduced noise-robust speech processing tasks with a small number of microphones [4], [44].…”