Speech Recognition: What’s Left? Dr Michael Picheny 12 November 2019

Duration: 59 mins 29 secs

Share this media item:

Embed this media item:

<iframe width="_width_" height="_height_" src="https://sms.cam.ac.uk/media/3108959/embed" frameborder="0" scrolling="no" allowfullscreen></iframe>

Choose size:

About this item

Available Formats

About this item

Description:	This talk examines speech recognition issue, comparing and contrasting them to what is known about human perception. With recent advances in Deep Learning, it is suggested that it is now achievable for Word Error Rates to be comparable to human listeners. This talk specifically highlights issues with accented, noisy speech, different speaking styles, multilingual speech recognition and more. And through demonstrations in comparison to human perception, there is still significant work in speech recognition research from the community.

Description:

This talk examines speech recognition issue, comparing and contrasting them to what is known about human perception. With recent advances in Deep Learning, it is suggested that it is now achievable for Word Error Rates to be comparable to human listeners. This talk specifically highlights issues with accented, noisy speech, different speaking styles, multilingual speech recognition and more. And through demonstrations in comparison to human perception, there is still significant work in speech recognition research from the community.


Created:	2019-11-25 13:43
Collection:	Information Engineering Distinguished Lecture Series
Publisher:	University of Cambridge
Copyright:	Dr Michael Picheny
Language:	eng (English)


Abstract:	Recent speech recognition advances on the SWITCHBOARD corpus suggest that because of recent advances in Deep Learning, we now achieve Word Error Rates comparable to human listeners. Does this mean the speech recognition problem is solved and the community can move on to a different set of problems? In this talk, we examine speech recognition issues that still plague the community and compare and contrast them to what is known about human perception. We specifically highlight issues in accented speech, noisy/reverberant speech, speaking style, rapid adaptation to new domains, and multilingual speech recognition. We try to demonstrate that compared to human perception, there is still much room for improvement, so significant work in speech recognition research is still required from the community.

Abstract:

Recent speech recognition advances on the SWITCHBOARD corpus suggest that because of recent advances in Deep Learning, we now achieve Word Error Rates comparable to human listeners. Does this mean the speech recognition problem is solved and the community can move on to a different set of problems? In this talk, we examine speech recognition issues that still plague the community and compare and contrast them to what is known about human perception. We specifically highlight issues in accented speech, noisy/reverberant speech, speaking style, rapid adaptation to new domains, and multilingual speech recognition. We try to demonstrate that compared to human perception, there is still much room for improvement, so significant work in speech recognition research is still required from the community.

Available Formats

Format	Quality	Bitrate	Size
MPEG-4 Video	1280x720	2.99 Mbits/sec	1.30 GB	View	Download
MPEG-4 Video	640x360	1.93 Mbits/sec	864.45 MB	View	Download
WebM	1280x720	2.35 Mbits/sec	1.02 GB	View	Download
WebM	640x360	406.25 kbits/sec	177.04 MB	View	Download
iPod Video	480x270	520.17 kbits/sec	226.63 MB	View	Download
MP3	44100 Hz	249.75 kbits/sec	108.93 MB	Listen	Download
Auto *	(Allows browser to choose a format it supports)