Effect of speech priors in single-channel speech-music separation for ASR
13th Annual Conference of the International Speech Communication Association 2012, INTERSPEECH 2012, Portland, OR, Amerika Birleşik Devletleri, 9 - 13 Eylül 2012, cilt.2, ss.1234-1237, (Tam Metin Bildiri)
- Yayın Türü: Bildiri / Tam Metin Bildiri
- Cilt numarası: 2
- Basıldığı Şehir: Portland, OR
- Basıldığı Ülke: Amerika Birleşik Devletleri
- Sayfa Sayıları: ss.1234-1237
- Anahtar Kelimeler: Prior speech model, Speech recognition, Speech-music separation
- Boğaziçi Üniversitesi Adresli: Evet
Özet
In this study, we extend the catalog-based single-channel speech-music separation method such that it incorporates prior speech information to enhance the separation performance of the method. We develop an inference method that enables us to use a speech prior model. The method uses a complex Gaussian observation model and an inverse-gamma prior model. We compare the separation performance of the catalog-based method with and without the prior speech model in both complex Gaussian and Poisson observation models. It is shown that for both observation models incorporating prior speech information improves the separation performance of the catalogbased method.