[AUDITORY] ICASSP 2021 Acoustic Echo Cancellation Challenge [Extended deadline] (Ross Cutler )

Subject: [AUDITORY] ICASSP 2021 Acoustic Echo Cancellation Challenge [Extended deadline] From: Ross Cutler <000001144701df79-dmarc-request@xxxxxxxx> Date: Tue, 11 Aug 2020 18:11:14 +0000 --_000_CO2PR00MB0183842AF8FE98A0B7755BAFEB451CO2PR00MB0183namp_ Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable Our team at Microsoft recently organized the Deep Noise Suppression Challen= ge at INTERSPEECH 2020<https://dns-challenge.azurewebsites.net/> to help st= imulate research in this important area of research. The response from that= challenge exceeded our expectations with 19 teams submitting 28 models for= noise suppression. The dataset and test framework continue to be cloned an= d forked at a good rate since the challenge, indicating it has good utility= for the research community. Another important area for speech enhancement is acoustic echo cancellation= , which is still a top issue in audio communication and conferencing system= s. We are submitting a proposal to organize an Acoustic Echo Cancellation C= hallenge at ICASSP 2021 to stimulate research in this area. The challenge w= ill have an open source dataset for acoustic echo cancellation and detectio= n, captured using a large-scale crowdsourcing effort. We will provide a dat= aset of 10,000 different real environments, audio devices, and human speake= rs in the following scenarios: 1. Far-end single talk, no echo path change 2. Far-end single talk, echo path change 3. Near-end single talk, no echo path change 4. Double talk, no echo path change 5. Double talk, echo path change 6. Sweep signal for RT60 estimation Each scenario includes the loudspeaker, microphone, and loopback signal. Be= sides the real data, we will also provide a synthetic dataset which is usef= ul for training, and a real anechoic chamber dataset of 1000 devices. The challenge is for real-time acoustic echo cancellation (AEC): * The AEC must run in less than T/2 (in ms) to process a frame of size = T (in ms) on an Intel Core i5 quad-core machine clocked at 2.4 GHz or equiv= alent processors. Frame length T must be less than or equal to 40ms (0<T<= =3D40ms). * The AEC method may have a maximum of 40ms look ahead, i.e., the algor= ithmic delay is 40ms. To infer the current frame T (in ms), the algorithm c= an access any number of past frames but only 40ms of future frames (T+40ms)= . * We will provide a base model implementation. * We will provide an objective and subjective evaluation framework. * The AEC can be a deep model, a traditional signal processing algorith= m, or a mix of the two. There are no restrictions on the AEC aside from the= run time and receptive field described above. * Prizes will be awarded to the top three participants. For the proposal, ICASSP requires us to provide them the list of potential = participants with their potential title and author(s) details. If you are i= nterested to participate in this challenge, please send ross.cutler@xxxxxxxx= ft.com<mailto:ross.cutler@xxxxxxxx> the following details by August 17= th. 1. A potential title of your paper. This can be updated later. 2. A potential list of authors. This can be updated later. 3. A webpage/URL of your personal webpage, LinkedIn page, or your resear= ch lab page. We look forward to hearing back from you and we hope to see you participate= in the challenge. Regards, Acoustic Echo Cancellation Challenge Organizers --_000_CO2PR00MB0183842AF8FE98A0B7755BAFEB451CO2PR00MB0183namp_ Content-Type: text/html; charset="us-ascii" Content-Transfer-Encoding: quoted-printable <html xmlns:v=3D"urn:schemas-microsoft-com:vml" xmlns:o=3D"urn:schemas-micr= osoft-com:office:office" xmlns:w=3D"urn:schemas-microsoft-com:office:word" = xmlns:dt=3D"uuid:C2F41010-65B3-11d1-A29F-00AA00C14882" xmlns:m=3D"http://sc= hemas.microsoft.com/office/2004/12/omml" xmlns=3D"http://www.w3.org/TR/REC-= html40"> <head> <meta http-equiv=3D"Content-Type" content=3D"text/html; charset=3Dus-ascii"= > <meta name=3D"Generator" content=3D"Microsoft Word 15 (filtered medium)"> <style></style> </head> <body lang=3D"EN-US" link=3D"#0563C1" vlink=3D"#954F72" style=3D"word-wrap:= break-word"> <div class=3D"WordSection1"> Our team at Microsoft recently organized the <a href= =3D"https://dns-challenge.azurewebsites.net/"> Deep Noise Suppression Challenge at INTERSPEECH 2020</a> to help stimulate = research in this important area of research. The response from that challen= ge exceeded our expectations with 19 teams submitting 28 models for noise s= uppression. The dataset and test framework continue to be cloned and forked at a good rate since the challe= nge, indicating it has good utility for the research community.<o:p></o:p><= /p> Another important area for speech enhancement is aco= ustic echo cancellation, which is still a top issue in audio communication = and conferencing systems. We are submitting a proposal to organize an Acous= tic Echo Cancellation Challenge at ICASSP 2021 to stimulate research in this area. The challenge will have an= open source dataset for acoustic echo cancellation and detection, captured= using a large-scale crowdsourcing effort. We will provide a dataset of 10,= 000 different real environments, audio devices, and human speakers in the following scenarios:<o:p></o:p></= p> <ol style=3D"margin-top:0in" start=3D"1" type=3D"1"> <li class=3D"MsoListParagraphCxSpFirst" style=3D"margin-left:0in;mso-add-sp= ace:auto;mso-list:l2 level1 lfo1"> Far-end single talk, no echo path change<o:p></o:p></li><li class=3D"MsoLis= tParagraphCxSpMiddle" style=3D"margin-left:0in;mso-add-space:auto;mso-list:= l2 level1 lfo1"> Far-end single talk, echo path change<o:p></o:p></li><li class=3D"MsoListPa= ragraphCxSpMiddle" style=3D"margin-left:0in;mso-add-space:auto;mso-list:l2 = level1 lfo1"> Near-end single talk, no echo path change<o:p></o:p></li><li class=3D"MsoLi= stParagraphCxSpMiddle" style=3D"margin-left:0in;mso-add-space:auto;mso-list= :l2 level1 lfo1"> Double talk, no echo path change<o:p></o:p></li><li class=3D"MsoListParagra= phCxSpMiddle" style=3D"margin-left:0in;mso-add-space:auto;mso-list:l2 level= 1 lfo1"> Double talk, echo path change<o:p></o:p></li><li class=3D"MsoListParagraphC= xSpLast" style=3D"margin-left:0in;mso-add-space:auto;mso-list:l2 level1 lfo= 1"> Sweep signal for RT60 estimation<o:p></o:p></li></ol> Each scenario includes the loudspeaker, microphone, = and loopback signal. Besides the real data, we will also provide a syntheti= c dataset which is useful for training, and a real anechoic chamber dataset= of 1000 devices.<o:p></o:p> The challenge is for real-time acoustic echo cancell= ation (AEC): <o:p></o:p> <ul style=3D"margin-top:0in" type=3D"disc"> <li class=3D"MsoListParagraphCxSpFirst" style=3D"margin-left:0in;mso-add-sp= ace:auto;mso-list:l0 level1 lfo2"> The AEC must run in less than T/2 (in ms) to process a frame of size T (in = ms) on an Intel Core i5 quad-core machine clocked at 2.4 GHz or equivalent = processors. Frame length T must be less than or equal to 40ms (0<T<= =3D40ms).<o:p></o:p></li><li class=3D"MsoListParagraphCxSpMiddle" style=3D"= margin-left:0in;mso-add-space:auto;mso-list:l0 level1 lfo2"> The AEC method may have a maximum of 40ms look ahead, i.e., the algorithmic= delay is 40ms. To infer the current frame T (in ms), the algorithm can acc= ess any number of past frames but only 40ms of future frames (T+40ms).<o:p>= </o:p></li><li class=3D"MsoListParagraphCxSpMiddle" style=3D"margin-left:0i= n;mso-add-space:auto;mso-list:l0 level1 lfo2"> We will provide a base model implementation.<o:p></o:p></li><li class=3D"Ms= oListParagraphCxSpMiddle" style=3D"margin-left:0in;mso-add-space:auto;mso-l= ist:l0 level1 lfo2"> We will provide an objective and subjective evaluation framework.<o:p></o:p= ></li><li class=3D"MsoListParagraphCxSpMiddle" style=3D"margin-left:0in;mso= -add-space:auto;mso-list:l0 level1 lfo2"> The AEC can be a deep model, a traditional signal processing algorithm, or = a mix of the two. There are no restrictions on the AEC aside from the run t= ime and receptive field described above.<o:p></o:p></li><li class=3D"MsoLis= tParagraphCxSpLast" style=3D"margin-left:0in;mso-add-space:auto;mso-list:l0= level1 lfo2"> Prizes will be awarded to the top three participants.<o:p></o:p></li></ul> For the proposal, ICASSP requires us to provide them= the list of potential participants with their potential title and author(s= ) details. If you are interested to participate in this challenge, please s= end <a href=3D"mailto:ross.cutler@xxxxxxxx">ross.cutler@xxxxxxxx</= a> the following details by August 17th.<o:p></o:p> <ol style=3D"margin-top:0in" start=3D"1" type=3D"1"> <li class=3D"MsoListParagraphCxSpFirst" style=3D"margin-left:0in;mso-add-sp= ace:auto;mso-list:l1 level1 lfo3"> A potential title of your paper. This can be updated later.<o:p></o:p></li>= <li class=3D"MsoListParagraphCxSpMiddle" style=3D"margin-left:0in;mso-add-s= pace:auto;mso-list:l1 level1 lfo3"> A potential list of authors. This can be updated later.<o:p></o:p></li><li = class=3D"MsoListParagraphCxSpLast" style=3D"margin-left:0in;mso-add-space:a= uto;mso-list:l1 level1 lfo3"> A webpage/URL of your personal webpage, LinkedIn page, or your research lab= page.<o:p></o:p></li></ol> We look forward to hearing back from you and we hope= to see you participate in the challenge.<o:p></o:p> Regards,<o:p></o:p> Acoustic Echo Cancellation Challenge Organizers<o:p>= </o:p> <o:p> </o:p> </div> </body> </html> --_000_CO2PR00MB0183842AF8FE98A0B7755BAFEB451CO2PR00MB0183namp_--

This message came from the mail archive
src/postings/2020/
maintained by:

DAn Ellis <dpwe@ee.columbia.edu>
Electrical Engineering Dept., Columbia University