[Subject Prev][Subject Next][Thread Prev][Thread Next][Date Index][Thread Index]

[hts-users:03940] Re:


On 2013-11-20 05:31, Lilly Christina wrote:
I have few basic doubts in HTS.
In HTS demo, what is the significance of the raw files being used in
48kHz sampling rate?

It has been shown that sampling rates higher than the "usual" 16 kHz can improve speaker similarity in synthetic voices, see this paper:
http://dx.doi.org/10.1109/ICASSP.2010.5495562

The experiments in the paper actually use 44.1 kHz, but this somewhat odd number stems from compact discs, which is not relevant for us I guess (see http://en.wikipedia.org/wiki/44,100_Hz for a nice explanation).

Best, Dietmar

--
----------------------------------------------------------------------
Dipl.-Ing. Dietmar Schabus | Researcher
phone +43 1 5052830-48 | fax -99 | schabus@xxxxxx | http://userver.ftw.at/~schabus/

FTW Telecommunications Research Center Vienna
Donau-City-Straße 1/3 | 1220 Vienna | Austria | www.ftw.at

Follow-Ups
[hts-users:03941] Re:, Simon King
References
[hts-users:03938], Lilly Christina