Hi, all,
The reply from Lyra:
The codec has been trained on thousands of hours of speech in more than 70
languages and tested on many different speakers and accents. It relies only on
log-mel spectrograms per frame and no long-term speaker id. For some audio
samples you can check out
Al,
My understanding ii that Lyra is currenlty optimised for arm64 systems.
Like LPCNet, it may still need a port of several key functions to some
sort of x86 SIMD instruction set like AVX.
- David
On 9/4/21 8:34 pm, Al Beard via Freetel-codec2 wrote:
Hi all,
I've loaded this "bazel"
I've been making some progress on open source, HF data modes:
http://www.rowetel.com/?p=7665
Currenlty having fun sending bursts of data frames to KiwiSDRs around
Australia.
- David
___
Freetel-codec2 mailing list