1 |
What Makes a Speaker Recognizable in TV Broadcast? Going Beyond Speaker Identification Error Rate
|
|
|
|
In: Interspeech 2015 ; ERRARE Workshop, a satellite event of Interspeech 2015. ; https://hal.archives-ouvertes.fr/hal-01433205 ; ERRARE Workshop, a satellite event of Interspeech 2015., 2015, Sinaia, Romania (2015)
|
|
Abstract:
International audience ; Speaker identification approaches for TV broadcast are usually evaluated and compared based on global error rates derived from the overall duration of missed detection, false alarm and confusion. Based on the analysis of the output of the systems submitted to the final round of the French evaluation campaign REPERE, this paper highlights the fact that these average met-rics lead to the incorrect intuition that current state-of-the-art algorithms partially recognize all speakers. Setting aside incorrect diarization and adverse acoustic conditions, we show that their performance is in fact essentially bi-modal: in a given show, either all speech turns of a speaker are correctly identified or none of them are. We then proceed with trying to understand and explain this behavior, through perfomance prediction experiments. These experiments show that the most discriminant speaker characteristics are – first – their total speech duration in the current show and – then only – the amount of training data available to build their acoustic model.
|
|
Keyword:
[INFO.INFO-CL]Computer Science [cs]/Computation and Language [cs.CL]; error analysis; speaker recognition; TV broadcast
|
|
URL: https://hal.archives-ouvertes.fr/hal-01433205/document https://hal.archives-ouvertes.fr/hal-01433205 https://hal.archives-ouvertes.fr/hal-01433205/file/Charlet2015.pdf
|
|
BASE
|
|
Hide details
|
|
2 |
Automatic Detection of Phoneme-Based Anomalies in Dysarthric Speech
|
|
|
|
In: ISSN: 1936-7228 ; EISSN: 1936-7236 ; ACM Transactions on Accessible Computing ; https://hal.archives-ouvertes.fr/hal-01485312 ; ACM Transactions on Accessible Computing , ACM New York, NY, USA 2015, Vol. 6 n° 3 (2015)
|
|
BASE
|
|
Show details
|
|
3 |
Automatic speech processing for dysarthria: A study of Inter-pathology variability
|
|
|
|
In: 18th International Congress of Phonetic Sciences 18 ; https://hal.archives-ouvertes.fr/hal-01498821 ; 18th International Congress of Phonetic Sciences 18, Aug 2015, Glasgow, United Kingdom. pp.5 (2015)
|
|
BASE
|
|
Show details
|
|
4 |
Traitement automatique de la parole dysarthrique: Etude de la variabilité inter-pathologique
|
|
|
|
In: Journées de Phonétique Clinique 6 ; https://hal.archives-ouvertes.fr/hal-01498892 ; Journées de Phonétique Clinique 6, Jun 2015, Montpellier, France. non paginé (2015)
|
|
BASE
|
|
Show details
|
|
5 |
Détection automatique d'anomalies dans la parole dysarthrique
|
|
|
|
In: Journées de Phonétique Clinique 6 ; https://hal.archives-ouvertes.fr/hal-01498824 ; Journées de Phonétique Clinique 6, Jun 2015, Montpellier, France. non paginé (2015)
|
|
BASE
|
|
Show details
|
|
|
|