Hi, I'm trying to build an lsp codebook to learn about the
process, and I'm afraid that the vq hints
http://www.xiph.org/archives/vorbis-dev/0000.html
talk rather about residual training.
So I can now produce .vqi files and add to them with lspvqtrain,
but can anyone supply hints for the rest of the procedure?
vqbuild will only produce a "cloud"-type quantlist, and
sequencep is pretty much ignored.
But if you use it anyway, you can go by hand through the xxx.vqh
file and guesstimate enough to make a bogus .vql file.
Then I ran latticebuild to get a first try .vqh file
Any comments about what's allowed and how to choose addmul?
Then I tried latticetune .vqh-file .vqd-file
ohoh tha name had to be precisely latticetune (or restune?)
ok I tried restune to read multiple cols. from lsp0coeff_0.vqd
still not a big difference. Tried changing values in my .vql
file. This made some difference. but I can't figure it out.
How does one choose reasonable value0..valuen on the second input
line to latticebuild? I would have thought that the lspvqtrain
output (xxx.vqi) had this info in some ready-to-use fashion.
Thanks, Erik.
--- >8 ----
List archives: http://www.xiph.org/archives/
Ogg project homepage: http://www.xiph.org/ogg/
To unsubscribe from this list, send a message to
'vorbis-dev-request@xiph.org'
containing only the word 'unsubscribe' in the body. No subject is
needed.
Unsubscribe messages sent to the list will be ignored/filtered.
