Detail publikace

Application of Re-segmentation in Very Low Bit Rate Speech Coding

Originální název

Application of Re-segmentation in Very Low Bit Rate Speech Coding

Anglický název

Application of Re-segmentation in Very Low Bit Rate Speech Coding

Jazyk

en

Originální abstrakt

The aim of our effort is to reach higher quality of resulting speech coded by very low bit rate (VLBR) segmental coder. The basic units are found automatically in a training database using temporal decomposition and vector quantization. They are modeled by HMMs. Then two methods of re-segmentation were used in order to find new longer units. In the first approach borders are set to the centers of previous units. In the second, borders are fixed to the centers of middle HMM states of previous units. Number of frames in new units is conditioned to be bigger than a fixed constant. Hence, new units can consist of a several previous segments. Decreasing transition noise of resultant speech was obtained using these techniques.

Anglický abstrakt

The aim of our effort is to reach higher quality of resulting speech coded by very low bit rate (VLBR) segmental coder. The basic units are found automatically in a training database using temporal decomposition and vector quantization. They are modeled by HMMs. Then two methods of re-segmentation were used in order to find new longer units. In the first approach borders are set to the centers of previous units. In the second, borders are fixed to the centers of middle HMM states of previous units. Number of frames in new units is conditioned to be bigger than a fixed constant. Hence, new units can consist of a several previous segments. Decreasing transition noise of resultant speech was obtained using these techniques.

BibTex


@inproceedings{BUT14399,
  author="Petr {Motlíček}",
  title="Application of Re-segmentation in Very Low Bit Rate Speech Coding",
  annote="The aim of our effort is to reach higher quality of resulting speech coded by very
low bit rate (VLBR) segmental coder. The basic units are found automatically in a training
database using temporal decomposition and vector quantization. They are modeled by HMMs.
Then two methods of re-segmentation were used in order to find new longer units.
In the first approach borders are set to the centers of previous units. In the second,
borders are fixed to the centers of middle HMM states of previous units.
Number of frames in new units is conditioned to be bigger than a fixed constant. Hence,
new units can consist of a several previous segments. Decreasing transition noise of resultant
speech was obtained using these techniques.",
  address="Faculty of Electrical Engineering and Communication BUT",
  booktitle="Proceedings of 7th Conference STUDENT EEICT 2001",
  chapter="14399",
  institution="Faculty of Electrical Engineering and Communication BUT",
  year="2001",
  month="march",
  pages="269--274",
  publisher="Faculty of Electrical Engineering and Communication BUT",
  type="conference paper"
}