Hello,
I am trying to use speex for speech encoding on my ARM based device.
I am facing issues with quality of voice with the narrow band configuration
(8k sampling rate).
With complexity set to 3 and VBR enabled, I am getting high bitrate output.
As an example,
My input is 8K stereo and I am passing the encoder with 160 samples at a
time as per the manual and samples.
When there is a speech I get between 40 to 92 bytes of encoded date out of
these 160 samples.
As I am using very low bandwith wireless network for sending encoded speech
data to remote location, this size is not acceptable to me as there is a lot
of packet drops on network at this rate.
If I change the configuration parameters with max VBR limit at 40000 and
complexity set to 10 then I get output upto 30 bytes. But in this case the
quality becomes very poor.
Can anybody help me what could be optimum speex setting for 8K stereo to
generate lowest bitrate with better voice quality?
Any guidance or any help is appreciate.
regards,
Ketan
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
http://lists.xiph.org/pipermail/speex-dev/attachments/20110106/c7ee0cb5/attachment.htm
