-- MikeLincoln - 30 Jan 2007

Schedule Autumn 2006

26 Sep 2006

Introductory meeting followed by report from Steve R. and or Maria W. on the recent 'Preparing for FP7' event (and - as requested by Gabriel - some informal reports back from Interspeech)

3 Oct 2006

Simon K. - Report on work done at Johns Hopkins, summer 2006

• WS06AFSR_summary_for_CSTR.ppt

10 Oct 2006

Tom Griffiths (ANC)

External Speaker

Human Activity Recognition Using Objects

The computer vision endeavour of activity recognition from video has wide-ranging potential applications, from the monitoring of elderly people at home to shopping centre security systems. In this talk I'll describe the approach we're taking to this task: inferring a person's activity from the way they use objects in the environment. I'll introduce how we break the problem down in to three levels (object tracking, action primitive extraction, and high-level activity modelling), give an overview of our approach to each one (with demos in some cases), and explain some future directions.

17 Oct 2006

Korin R.

A Trajectory Mixture Density Network for the Acoustic-articulatory Inversion Mapping

This talk presents work which appeared at the recent Interspeech 2006 conference. A trajectory model is proposed which is based on a mixture density network trained with target features augmented with dynamic features. This is then combined with an algorithm for estimating maximum likelihood trajectories which respects constraints between the static and derived dynamic features. I evaluated this model on an inversion mapping task. The introduction of the trajectory model successfully reduced root mean square error by up to 7.5%, as well as increasing correlation scores.

24 Oct 2006

Giulia G.

31 Oct 2006

Brainstorming

Simon K., Partha L., Mirjam W. and Joe F.

• Mirjam: on previous multilingual work by othersmultiling.pdf
• Partha: on Global Phone and other data consideredbrain_storm.pdf
• Simon: recap of the JHU Tandem-AF results WS06AFSR_summary_for_CSTR.ppt
• Partha: on his proposed thesis topicbrain_storm.pdf
• Joe: on learning ANN output classes with balanced priors slides.pdf
• Simon: on graphemes and other ideas from LIPS graphemes_etc.pdf

7 Nov 2006

14 Nov 2006

Tim Hospedales (IPAB)

External Speaker

21 Nov 2006

Joe F.

Multichannel dialogue act decoding of multiparty meetings

A number of authors have considered the task of dialogue act (DA) decoding for multiparty meetings, suggesting that information about the structure of the interaction (i.e. the sequence of statements, questions, interuptions etc) should prove useful in subsequent processing steps.

Previous approaches have typically considered each channel (speaker) in isolation when performing inference.

In this talk I present a framework for performing full joint classification and segmentation of all channels together. I take a multi-stream modelling approach using dynamic Bayesian networks (DBNs), with a state stream for each channel, and observations in the form of virtual evidence (VE), which consists of posteriors from conditional random fields (CRFs).

This is work in progress, and arose from discussions with Matthias Zimmermann after this year's MLMI.

* slides.pdf: Talk slides

28 Nov 2006

Brainstorming

Leader required!

5 Dec 2006

Mike L

12 Dec 2006

Schedule Spring 2007

23 Jan 2007

Leonardo B. - Loquendo's TTS system

30 Jan 2007

Fiona

06 Feb 2007

No CSTR meeting due to :

INSTITUTE FOR ADAPTIVE & NEURAL COMPUTATION SEMINAR

Yee Whye Teh Department of Computer Science School of Computing National University of Singapore

Tuesday, 6th February, 11am D-Floor Lecture Theatre (Jim Howe Room) 5 Forrest Hill

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Stick-breaking Construction for the Indian Buffet Process

The Indian buffet Process (IBP) is a recently proposed latent feature model where each object is modelled using a potentially unbounded number of binary latent features. It has had a variety of applications, including matrix factorization, causal inference, and psychological choice modelling. However, due to the unbounded nature of the model, standard Markov chain Monte Carlo inference techniques like Gibbs sampling is cumbersome and inefficient in IBPs. In this talk, I will reformulate the IBP model using a stick-breaking construction, and show that this leads to straightforward and efficient MCMC inference for the IBP. Furthermore, we will see that there are interesting and strong connections between the stick-breaking construction for the IBP, and the standard stick-breaking construction for the more well-known Dirichlet process.

There is another talk by the same speaker at 3.30pm in the same venue: A Bayesian Interpretation of Interpolated Kneser-Ney

13 Feb. 2007

20 Feb. 2007

27 Feb. 2007

Brain Storming

06 Mar. 2007

13 Mar. 2007

No Meeting

20 Mar. 2007

Craig Nicol Using Timbre Spaces to explore audio

27 Mar. 2007

No Meeting

03 Apr. 2007

No Meeting

10 Apr. 2007

Mirjam

17 Apr. 2007

Brain Storming - Simon and Steve will talk about a project proposal they are writing called "EMIME: Efficient Multilingual Interaction in Mobile Environments" which involves ASR , HMM-based TTS, speaker- and language-adaptation, all glued together with machine translation.

24 Apr. 2007

No Meeting

01 May 2007

Maria Walters

08 May 2007

Songfang Huang - Where We Are Towards A Multimodal LM: A Progress Report

15 May 2007

Sebastian Moeller, Deutsche Telekom

22 May 2007

UNIVERSITY HOLIDAY

29 May 2007

Brain Storming

05 June 2007

Partha Lal

12 June 2007

No Meeting

19 June 2007

No Meeting

26 June 2007

No Meeting

Summer Break

04 Sept. 2007

Steve Renals

11 Sept. 2007

No Meeting

18 Sept. 2007

Sebastien Anderson

25 Sept. 2007

Brainstorming

02 Oct. 2007

No Talk

09 Oct. 2007

Hiroshi Shimodara

16 Oct. 2007

Vipperla Ravi Chander

23 Oct. 2007

Zhang Le

30 Oct. 2007

Junichi Yamagishi

06 Nov. 2007

Joao Cabral

13 Nov. 2007

Brain Storming

20 Nov. 2007

Giulia Garau

27 Nov. 2007

Chao Qin

04 Dec. 2007

Alfred Dielmann

11 Dec. 2007

Sebastien Anderson

18 Dec. 2007

Vipperla Ravi Chander

Schedule Spring 2008

22 Jan 2008

Korin Richmond

29 Jan 2008

Brain Storming

05 Feb 2008

Maria Wolters

12 Feb. 2008

Zhenhua Ling

19 Feb. 2008

Ivan Himawan

26 Feb. 2008

Brain Storming (Steve - CUES and SSPnet proposals)

05 Mar. 2008 [UNUSUAL DAY - Wednesday. UNUSUAL TIME - 4pm. VENUE - conference suite]

Simon King

11 Mar. 2008

Yi-Jian Wu (Nagoya Institute of Technology) - Minimum generation error training for HMM-based speech synthesis

18 Mar. 2008

Rob Clark

25 Mar. 2008

Ivan Himawan

01 Apr. 2008

Jochen Ehnes

08 Apr. 2008

no meeting

15 Apr. 2008

External ?

22 Apr. 2008

Mike Lincoln

29 April 2008

Brain Storming (Steve)

06 May 2008

Peter Bell - POSTPONED

13 May 2008

Michael Berger

20 May 2008

UNIVERSITY HOLIDAY

27 May 2008

no meeting

03 June 2008

Tanja Kocjancic

10 June 2008

Martin Tietze: thesis proposal - POSTPONED

17 June 2008 [UNUSUAL VENUE: Seminar room]

Volker Strom

24 June 2008

No meeting

Start of Summer Break

29 July 2008

Tomoki Toda (Nara Advanced Institute of Science and Technology - NAIST) - Trajectory HMMs

End of Summer Break

02 Sept. 2008

no meeting

09 Sept. 2008

SPARE SLOT [was: Partha Lal (NB MLMI is taking place this week)]

16 Sept. 2008

OPEN SLOT

23 Sept. 2008

No meeting: Interspeech is taking place this week

30 Sept. 2008

no meeting

07 0ct. 2008

'What to do in the meeting' meeting

13 Oct. 2008

No Meeting

20 Oct. 2008

Dong Wang

27 Oct. 2008

No meeting?

03 Nov. 2008

CSTR General Meeting

10 Nov. 2008

John Niekrasz

17 Nov. 2008

???

24 Nov. 2008

???

Friday 28 Nov. 2008 (13:30, 4.31/33)

Zhang Le - pre-viva talk

01 Dec. 2008

Steve Renals - Overview of AMIDA

08 Dec. 2008

No meeting

15 Dec. 2008

???

Friday 19 Dec. 2008

11:00 Gregor Hofer DDD

14:00 AMIDA demos in the IMR

Christmas Break

12 Jan. 2009

No meeting

19 Jan. 2009

Songfang Huang (DDD)

26 Jan. 2009

Joao Cabral (DDD)

2 Feb. 2009

Tim Mills

9 Feb. 2009

Michael Berger (Thesis Proposal)

16 Feb. 2009

Peter Bell (DDD)

(was Junichi Yamagashi - needs to be moved later)

23 Feb. 2009

Oliver Watts (Thesis Proposal) UNUSUAL TIME: 11am

(was Matthew Aylett - needs to be moved later)

2 Mar. 2009

No Meeting (room in use)

9 Mar. 2009

Leonardo Badino

16 Mar 2009

Sebastian Andersson

23 Mar. 2009

30 Mar. 2009

Sharon Goldwater

6 Apr. 2009

Dong Wang (DDD)

**Unusual time (2:00 pm)

Easter break

13 Apr. 2009

Maria Wolters

20 Apr. 2009

CSTR general Meeting

27 Apr. 2009

Rob Clark

4 May. 2009

Korin Richmond

11 May 2009

Jochen Ehnees

18 May 2009

No Meeting (Victoria Day)

25 May 2009

No Meeting

1 June 2009

Graham Pullin (University of Dundee)

8 June 2009

Tanja Kocjancic

15 June 2009

No meeting

22 June 2009

Volker Strom

29 June 2009

Martin Tietze

Summer Break

19th October 2009

Erich Zwyssig

"Digital Microphone Array - Design, Implementation and Speech Recognition Experiments"

EZ - 23thOct2009 - Presentation can now be found under attachments, see bottom of page ...

26th October 2009

EMIME (Simon King / Junichi Yamagishi)

Voice Cloning

2nd November 2009

No Meeting

9th November 2009

Tim Mills

"Wave of the future? Assessing the promise of 'reassigned' spectrograms" (postponed to 23 November)

16th November 2009

Partha Lal

Adaptation of Decision Tree Acoustic Models for Speech Recognition

23rd November 2009

Tim Mills

"Wave of the future? Assessing the promise of 'reassigned' spectrograms"

(Papers attached below. Also see here for Matlab scripts implementing the reassigned spectrogram.)

30th November 2009

No meeting

7th December 2009

Sharon Goldwater

14th December 2009

Mike Lincoln

Realtime ASR and Automatic Content Linking

Christmas Break

11th January 2010

(TBC) Matthew Aylett

18th January 2010

Mirjam Wester

25th January 2010

Keith Edwards

Commercialisation

1st February 2010

Adriana Stan

Romanian HTS

8th February 2010

No Meeting

15th February 2010

No Meeting

22nd February 2010

Theresa Wilson

1st March 2010

Sebastian Andersson - DDD

(speaker and topic subject to change)

8th March 2010

15th March 2010

22nd March 2010

Oliver Watts

29th March 2010

Liang Lu

5th April 2010

Optional Holiday (Easter Monday)

12th April 2010

Maria Wolters (MultiMemoHome )

19th April 2010

Optional Holiday ('Spring')

26th April 2010

No meeting

3rd May 2010

Michael Berger

10th May 2010

Michael Akeroyd, MRC Institute of Hearing Research (099_maa_edinburgh_may2010_draft1.pptx: slides)

17th May 2010

Holiday (Victoria Day #1)

24th May 2010

No Meeting

31st May 2010

No Meeting

7th June 2010

No Meeting

14th June 2010

No Meeting

21st June 2010

No Meeting (moved to 8th July)

28th June 2010

Jochen Ehnes

8th July 2010

Korin Richmond ( Note this talk is scheduled for 4-5pm on a Thursday!)

Summer Break

4th October 2010*

Junichi Yamagishi

22nd November 2010*

Matthew Aylett - Cereproc

6th December 2010

Erich Zwyssig

The aim of this one-week Autumn School (London, UK, 1-5 November 2010) is to train PhD students and researchers in the software development skills required to build reliable research software quickly and with a minimum of effort, and so maximize the impact of their research.

The Autumn School is presented by Dr. Greg Wilson in collaboration with SoundSoftware .ac.uk: Sustainable Software for Audio and Music Research.

Reading:

• vandehalle_SP09_reprod_research.pdf

• yale_2010_reproducible_research.pdf

• dubois_CSE05_maintaining_correctness_in_scien_programs.pdf

• cise-swc-2006.pdf

Code:

• code.tar.gz

Notes & Handout:

• SW-Carpentry.notes.pdf
• SW-Carpentry.handout.pdf

21st February 2011

Partha Lal

28th February 2011

Liang Lu

7th March 2011 (* PLEASE NOTE TIME: 10.00am *)

Sebastian Andersson Pre-Viva Talk.

14th March 2011

No meeting

21th March 2011

Christophe Veaux

28th March 2011

Mirjam Wester

4th April 2011

No meeting

11th April 2011

David Braude

18th April 2011

No meeting

25th April 2011

No Meeting

2nd May 2011

Martin Tietze

9th May 2011

Mark Sinclair

16th May 2011

No Talk

23th May 2011

No Talk

30th May 2011

Hiroyuki Segi (NHK, Japan Broadcasting Corporation)

6th June 2011

Oliver Watts