Abstract
The speech transmission index (STI) of a listening position within a given room indicates the quality and intelligibility of speech uttered in that room. The measure is very reliable for predicting speech intelligibility in many room conditions but requires an STI measurement of the impulse response for the room. We present a method for blindly estimating the STI without measuring or modeling the impulse response of the room using deep convolutional neural networks. Our model is trained entirely using simulated room impulse responses combined with clean speech examples from the DAPS dataset [1] and works directly on PCM audio. Our experiments show that our method predicts true STI with a high degree of accuracy-an average error of under 4%. It can also distinguish between different STI conditions to a level of granularity that is comparable to humans.
| Original language | English (US) |
|---|---|
| Title of host publication | 2018 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2018 - Proceedings |
| Publisher | Institute of Electrical and Electronics Engineers Inc. |
| Pages | 591-595 |
| Number of pages | 5 |
| ISBN (Print) | 9781538646588 |
| DOIs | |
| State | Published - Sep 10 2018 |
| Event | 2018 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2018 - Calgary, Canada Duration: Apr 15 2018 → Apr 20 2018 |
Publication series
| Name | ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings |
|---|---|
| Volume | 2018-April |
| ISSN (Print) | 1520-6149 |
Other
| Other | 2018 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2018 |
|---|---|
| Country | Canada |
| City | Calgary |
| Period | 4/15/18 → 4/20/18 |
Fingerprint
Keywords
- Speech enhancement
- Speech quality
- Speech transmission index
ASJC Scopus subject areas
- Software
- Signal Processing
- Electrical and Electronic Engineering
Cite this
}
Blind Estimation of the Speech Transmission Index for Speech Quality Prediction. / Seetharaman, Prem; Mysore, Gautham J.; Smaragdis, Paris; Pardo, Bryan A.
2018 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2018 - Proceedings. Institute of Electrical and Electronics Engineers Inc., 2018. p. 591-595 8461827 (ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings; Vol. 2018-April).Research output: Chapter in Book/Report/Conference proceeding › Conference contribution
TY - GEN
T1 - Blind Estimation of the Speech Transmission Index for Speech Quality Prediction
AU - Seetharaman, Prem
AU - Mysore, Gautham J.
AU - Smaragdis, Paris
AU - Pardo, Bryan A
PY - 2018/9/10
Y1 - 2018/9/10
N2 - The speech transmission index (STI) of a listening position within a given room indicates the quality and intelligibility of speech uttered in that room. The measure is very reliable for predicting speech intelligibility in many room conditions but requires an STI measurement of the impulse response for the room. We present a method for blindly estimating the STI without measuring or modeling the impulse response of the room using deep convolutional neural networks. Our model is trained entirely using simulated room impulse responses combined with clean speech examples from the DAPS dataset [1] and works directly on PCM audio. Our experiments show that our method predicts true STI with a high degree of accuracy-an average error of under 4%. It can also distinguish between different STI conditions to a level of granularity that is comparable to humans.
AB - The speech transmission index (STI) of a listening position within a given room indicates the quality and intelligibility of speech uttered in that room. The measure is very reliable for predicting speech intelligibility in many room conditions but requires an STI measurement of the impulse response for the room. We present a method for blindly estimating the STI without measuring or modeling the impulse response of the room using deep convolutional neural networks. Our model is trained entirely using simulated room impulse responses combined with clean speech examples from the DAPS dataset [1] and works directly on PCM audio. Our experiments show that our method predicts true STI with a high degree of accuracy-an average error of under 4%. It can also distinguish between different STI conditions to a level of granularity that is comparable to humans.
KW - Speech enhancement
KW - Speech quality
KW - Speech transmission index
UR - http://www.scopus.com/inward/record.url?scp=85054235329&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85054235329&partnerID=8YFLogxK
U2 - 10.1109/ICASSP.2018.8461827
DO - 10.1109/ICASSP.2018.8461827
M3 - Conference contribution
SN - 9781538646588
T3 - ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings
SP - 591
EP - 595
BT - 2018 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2018 - Proceedings
PB - Institute of Electrical and Electronics Engineers Inc.
ER -