Delcroix et al - REVERB Workshop 2014 LINEAR...

Info icon This preview shows pages 1–2. Sign up to view the full content.

View Full Document Right Arrow Icon
LINEAR PREDICTION-BASED DEREVERBERATION WITH ADVANCED SPEECH ENHANCEMENT AND RECOGNITION TECHNOLOGIES FOR THE REVERB CHALLENGE Marc Delcroix, Takuya Yoshioka, Atsunori Ogawa, Yotaro Kubo, Masakiyo Fujimoto, Nobutaka Ito, Keisuke Kinoshita, Miquel Espi, Takaaki Hori, Tomohiro Nakatani, Atsushi Nakamura NTT Communication Science Laboratories, NTT Corporation, Japan, [email protected] ABSTRACT This paper describes systems for the enhancement and recognition of distant speech recorded in reverberant rooms. Our speech enhance- ment (SE) system handles reverberation with blind deconvolution using linear filtering estimated by exploiting the temporal correla- tion of observed reverberant speech signals. Additional noise reduc- tion is then performed using an MVDR beamformer and advanced model-based SE. We employ this SE system as a front-end for our advanced automatic speech recognition (ASR) back-end, which uses deep neural network (DNN) based acoustic models and recurrent neural network based language models. Moreover, we ensure good interconnection between the SE front-end and ASR back-end using unsupervised model adaptation to reduce the mismatch caused by, for example, front-end processing artifacts. Our SE front-end greatly improves speech quality and achieves up to a 60 % relative word er- ror rate reduction for the real recordings of the REVERB challenge data, compared with a strong DNN-based ASR baseline. Index Terms Linear prediction-based dereverberation, model- based speech enhancement, DNN-based recognition. 1. INTRODUCTION The use of distant microphones to capture speech remains chal- lenging because noise and reverberation degrade the audible quality of speech and severely a ff ect the performance of automatic speech recognition (ASR). Much research has been undertaken to tackle the e ff ect of noise. However, dealing with reverberation has re- mained challenging because it has a long-term e ff ect that covers several analysis time frames, and it induces highly non-stationary distortions. Consequently, mitigating reverberation requires ded- icated approaches that exploit the long-term acoustic context and use e cient models of reverberation [1]. Such approaches di ff er fundamentally from conventional noise reduction techniques. This paper presents our contribution to the REVERB challenge for the enhancement and recognition of distant speech recorded in reverberant rooms [2]. The REVERB challenge data cover various reverberation conditions (reverberation times between 0.25 and 0.7 s) and also include a significant amount of noise. Dealing with such severe conditions requires powerful dereverberation and noise reduc- tion techniques. Our system combines speech enhancement (SE) techniques as a front-end to reduce reverberation and noise, and a state-of-the-art ASR back-end for optimal recognition performance.
Image of page 1

Info icon This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
Image of page 2
This is the end of the preview. Sign up to access the rest of the document.

{[ snackBarMessage ]}

What students are saying

  • Left Quote Icon

    As a current student on this bumpy collegiate pathway, I stumbled upon Course Hero, where I can find study resources for nearly all my courses, get online help from tutors 24/7, and even share my old projects, papers, and lecture notes with other students.

    Student Picture

    Kiran Temple University Fox School of Business ‘17, Course Hero Intern

  • Left Quote Icon

    I cannot even describe how much Course Hero helped me this summer. It’s truly become something I can always rely on and help me. In the end, I was not only able to survive summer classes, but I was able to thrive thanks to Course Hero.

    Student Picture

    Dana University of Pennsylvania ‘17, Course Hero Intern

  • Left Quote Icon

    The ability to access any university’s resources through Course Hero proved invaluable in my case. I was behind on Tulane coursework and actually used UCLA’s materials to help me move forward and get everything together on time.

    Student Picture

    Jill Tulane University ‘16, Course Hero Intern