Webpage for media files

Supplementary Contents (Audio examples) for IEEE Special Issue

Paper Submitted to Special Issue on Signal Models and Representations for IEEE Transactions on Audio, Speech and Language Processing

Fixed-Dimension Modified Sinusoidal Model: A High Quality Sparse Representation for Mixed Speech and Audio Signals

Pejman Mowlaee, Abolghasem Sayadiyan, and Hamid Sheikhzadeh,

Table 1. Sound quality assessment for male and female speakers.

Category\Description	Adult male	Old man	Old woman	Child	Adult female
Original signal	Male	oldman	old woman	Child	woman
FDMSM reconstructed using M=21	Male_M21	oldman_M21	oldwoman_M21	Child_M21	lgwc9_M21
FDMSM reconstructed using M=25	Male_M25	oldman_M25	oldwoman_M25	Child_M25	lgwc9_M25
FDMSM reconstructed using M=29	Male_M29	oldman_M29	oldwoman_M29	Child_M29	lgwc9_M29
FDMSM reconstructed using M=33	Male_M33	oldman_M33	oldwoman_M33	Child_M33	lgwc9_M33
FDMSM reconstructed using M=37	Male_M37	oldman_M37	oldwoman_M37	Child_M37	lgwc9_M37
FDMSM reconstructed using M=40	Male_M40	oldman_M60	oldwoman_M40	Child_M40	lgwc9_M40

Description of files: Male_M21 denote the FDMSM synthesized speech signal for a male speaker using Model order of M=21.

Table 3. Sound quality assessment for male and female speakers from Cooke database in [1].

Table 3(Test 1) Female	Table3 (Test 2) Male	Category
Test 1: “bbaj4n”	Test 2: “bbbe8n”	Description
lgwc9	bbbe8n	Original signal
lgwc9_M33	bbbe8n_M36	FDMSM reconstructed using M=37
lgwc9n_M40	bbbe8n_M40	FDMSM reconstructed using M=40
lgwc9n_M50	bbbe8n_M50	FDMSM reconstructed using M=50

Table 4. Sound quality assessment for signals of Quartet and Glockenspiel from SAQM database in [2].

Category	Table4 (Test 1)	Description	Table4 (Test 2)
Category	Quartet of 28 sec	Original signal	Glockenspiel of 35 sec
Original signal	quar_48	Original signal	gspi_35
FDMSM using M=25	FDMSM_quar48_1M25	FDMSM using M=20	FDMSM_gspi_35M20
FDMSM using M=29	FDMSM_quar48_1M29	FDMSM using M=24	FDMSM_gspi_35M24
FDMSM using M=33	FDMSM_quar48_1M33	FDMSM using M=25	FDMSM_gspi_35M25
FDMSM using M=37	FDMSM_quar48_1M37	FDMSM using M=28	FDMSM_gspi_35M28
FDMSM using M=40	FDMSM_quar48_1M40	FDMSM using M=37,40	FDMSM_gspi_35M37 FDMSM_gspi_35M40

Table 5. Sound quality assessment for mixtures.

No	Mixture Type	Description
1	Mix_bbae6p&bbaqzp	Mixture of Male and Female composed of bbae6p + bbaqzp
1	FDMSM_M40_Mix_bbae6p&bbaqzp	FDMSM reconstructed for the mixture of Female and Male M=33 composed bbae6p + bbaqzp

No	Mixture Type	Description
2	Mix_Child&Female	Mixture of Child and Female
2	FDMSM_Child&FemaleM33	FDMSM reconstructed for the mixture of Child and Female signal using M=33

No	Mixture Type	Description
3	Mix_bbaj4n&bbbe8n	Mixture of Female and Male composed of bbaj4n bbbe8n
	FDMSM_M33_Mix_bbaj4n&bbbe8n	FDMSM reconstructed for the mixture of Female and Male M=33 composed ofbbaj4n bbbe8n
	FDMSM_M50_Mix_bbaj4n&bbbe8n	FDMSM reconstructed for the mixture of Female and Male M=55 including bbaj4n bbbe8n

No	Mixture Type	Description
4	Mix_Song&Music	Original signal of mixture of Song and Music signals
	FDMSM_M40_Song&Music	FDMSM reconstructed for the mixture of Song and Music signals using M=40
	FDMSM_M60_Song&Music	FDMSM reconstructed for the mixture of Song and Music signals using M=60

Table 6. Sound quality assessment for corrupted signal with Babble/Harmonic noise at SNR=10dB.

Harmonic Noise
Description: harmonic noise, with a fundamental frequency of 300 Hz and its 10 harmonics
Male		Female		Mixture		Music
Male_harmonic	Original signal of harmonic noise and Male speaker	Female_harmonic	Original signal of harmonic noise and Female speaker	Mix_harmonic	Original signal of harmonic noise and speech mixture	Music_harmonic
Male_harmonic_M40	FDMSM reconstructed for the mixture of harmonic noise and Male speaker using M=40	Female_harmonic_M40	FDMSM reconstructed for the mixture of harmonic noise and Female speaker using M=40	Mix_harmonic_M37	FDMSM reconstructed for the mixture of harmonic noise and speech mixture using M=45	Music_harmonic_M45
Male_harmonic_M45	FDMSM reconstructed for the mixture of harmonic noise and Male speaker using M=45	Female_harmonic_M45	FDMSM reconstructed for the mixture of harmonic noise and Female speaker using M=45	Mix_harmonic_M40	FDMSM reconstructed for the mixture of harmonic noise and speech mixture using M=55	Music_harmonic_M55
Male_harmonic_M50	FDMSM reconstructed for the mixture of harmonic noise and Male speaker using M=50	Female_harmonic_M50	FDMSM reconstructed for the mixture of harmonic noise and Female speaker using M=50	Mix_harmonic_M45	FDMSM reconstructed for the mixture of harmonic noise and speech mixture using M=60	Music_harmonic_M60
Male_harmonic_M68	FDMSM reconstructed for the mixture of harmonic noise and Male speaker using M=68			Mix_harmonic_M50 Mix_harmonic_M60	FDMSM reconstructed for the mixture of harmonic noise and speech mixture using M=50,60

Babble Noise
Description: babble noise from NOISEX92
Male		Female		Mixture		Music
Male_babble	Original signal of babble noise and Male speaker	Female_babble	Original signal of babble noise and Female speaker	Mix_babble	Original signal of babble noise and speech mixture	Music_babble
Male_babble_M40	FDMSM reconstructed for the mixture of babble noise and Male speaker using M=40	Female_babble_M37	FDMSM reconstructed for the mixture of babble noise and Female speaker using M=40	Mix_babble_M40	FDMSM reconstructed for the mixture of babble noise and speech mixture using M=45	Music_babble_M34
Male_babbl_M45e	FDMSM reconstructed for the mixture of babble noise and Male speaker using M=45	Female_babbl_M40	FDMSM reconstructed for the mixture of babble noise and Female speaker using M=45	Mix_babble_M50	FDMSM reconstructed for the mixture of harmonic noise and speech mixture using M=55	Music_babble_M37
Male_babble_M50	FDMSM reconstructed for the mixture of babble noise and Male speaker using M=50	Female_babble_M50	FDMSM reconstructed for the mixture of babble noise and Female speaker using M=50	Mix_babble_M68	FDMSM reconstructed for the mixture of babble noise and speech mixture using M=60	Music_babble_M40
				Music_babble_M45 Music_babble_M50	FDMSM reconstructed for the mixture of babble noise and speech mixture using M=45,50

[1] The speech separation challenge database is downloadable from:

[Online]. Available: www.dcs.shef.ac.uk/~martin/SpeechSeparationChallenge.htm

[2] sound quality assessment material (SQAM) database downloadable from

[Online]. Available: http://andrew.csie.ncyu.edu.tw/html/mpeg4/sound.media.mit.edu/mpeg4/audio/sqam/index.html

References

P. Mowlaee, and A. Sayadiyan,
[1]. A Model Order based Sinusoidal representation for Audio Signals
Computer Systems and Applications, 2008. AICCSA 2008. IEEE/ACS International Conference on
March 31 2008-April 4 2008 Page(s):501 - 507

P. Mowlaee, and A. Sayadiyan,
[2]. Model-based Monaural Sound Separation by Split-VQ of sinusoidal parameter
16th European Signal Processing Conference EUSIPCO 2008
August 24 2008-August 29 2008 (Available online: http://www.eurasip.org/Proceedings/Eusipco2008/papers/1569096851.pdf)

P. Mowlaee, A. Sayadiyan, and M. Rahmati
[3]. Sparse Sinusoidal Representation for Speech and Music Signals
CCIS06 Springer Lecture Notes in Computer Science © Springer-Verlag Berlin Heidelberg 2008
Page(s):469 - 476, 2008 (Available online: http://www.springerlink.com/content/u572687721773842/)

P. Mowlaee, A. Sayadiyan, and M. Sheikhan
[4]. A Fixed Dimension Modified Sinusoid Model (FD-MSM) for Single Microphone Sound Separation
IEEE international Conference on Signal Processing and Communications, ICSPC 2007
Page(s):1183 - 1186, November 24-27, 2007.

Contacts

Emails

pmowlaee@ieee.org

pejmanmowlaee@gmail.com

P_Mowlaee@aut.ac.ir

web link:

http://ele.aut.ac.ir/pejmanmowlaee

Postal Address

Information Processing Laboratory
3rd floor, Aboureyhan Building
Amirkabir University of Technology

424, Hafez Ave., Tehran, Iran, 13597-45778

Cell: +98 9111834523, Fax: (+9821) 66406469

Back To my website