Re: [Freetel-codec2] Codec2 vs Lyra?

2021-04-14 Thread Random
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

Re: [Freetel-codec2] Codec2 vs Lyra?

2021-04-14 Thread David Rowe
f LPCNet. > > Best Regards. > > -- Ô­Ê¼Óʼþ -- > > *·¢¼þÈË:* "freetel-codec2" ; > *·¢ËÍʱ¼ä:* 2021Äê4ÔÂ7ÈÕ(ÐÇÆÚÈý) ÏÂÎç4:25 > *ÊÕ¼þÈË:* "freetel-codec2"; > > *Ö÷Ìâ:* Re: [Freetel-codec2] Codec2 vs Lyra? > > Hmm

Re: [Freetel-codec2] Codec2 vs Lyra?

2021-04-09 Thread Al Beard via Freetel-codec2
its > performance with that of LPCNet. > > Best Regards. > > -- Ô­Ê¼Óʼþ -- > > ·¢¼þÈË: "freetel-codec2" ; > ·¢ËÍʱ¼ä: 2021Äê4ÔÂ7ÈÕ(ÐÇÆÚÈý) ÏÂÎç4:25 > ÊÕ¼þÈË: "freetel-codec2"; > > Ö÷Ìâ: Re: [

[Freetel-codec2] ?????? Codec2 vs Lyra?

2021-04-08 Thread Random
Thanks a lot, David. Could you please squeeze Lyra to 2400bps ? Such that we can compare its performance with that of LPCNet. Best Regards. ---- ??:

Re: [Freetel-codec2] Codec2 vs Lyra?

2021-04-07 Thread Richard Shaw
On Wed, Apr 7, 2021 at 2:56 AM David Rowe wrote: > Hmm, I wonder how quickly I can get this into a FreeDV mode :-) > > 3000 bit/s is a bit high for HF, but nice for VHF. Might be cool for > the m17 project. > I may poke around building it for Fedora, but looks like I have to get the bazel

Re: [Freetel-codec2] Codec2 vs Lyra?

2021-04-07 Thread David Rowe
Hmm, I wonder how quickly I can get this into a FreeDV mode :-) 3000 bit/s is a bit high for HF, but nice for VHF.  Might be cool for the m17 project. On 7/4/21 5:18 pm, David Rowe wrote: Further to this, I've just bee told Google has open source Lyra :-)

Re: [Freetel-codec2] Codec2 vs Lyra?

2021-04-07 Thread David Rowe
Further to this, I've just bee told Google has open source Lyra :-) https://opensource.googleblog.com/2021/04/lyra-enabling-voice-calls-for-next-billion-users.html - David ___ Freetel-codec2 mailing list Freetel-codec2@lists.sourceforge.net

[Freetel-codec2] ?????? Codec2 vs Lyra?

2021-03-01 Thread Random
https://arxiv.org/abs/2102.09660 Here is the paper. ---- ??: "freetel-codec2"

[Freetel-codec2] ?????? Codec2 vs Lyra?

2021-03-01 Thread Random
I didn't find the paper link either. David, it seems that LPCnet fails to improve the MOS value, so there is no MOS in the LPCnet papers, while Lyra claims its superiority in terms of MOS. How large is the training dataset for Lyra ? ---- ??:

Re: [Freetel-codec2] Codec2 vs Lyra?

2021-02-28 Thread Adrian Musceac
Thanks Greg for the comments. I'm not a specialist, far from it. I've just seen ML applied in other domains and it did not inspire much confidence, however maybe this will be indeed different. If there is anyone who can pull this off, it is certainly Google. They do have access to a very large

Re: [Freetel-codec2] Codec2 vs Lyra?

2021-02-28 Thread Greg Maxwell
On Sun, Feb 28, 2021 at 12:41 PM Adrian Musceac wrote: > While interesting and newsworthy, I'd assume from the start that this codec > has the same advantages and pitfalls as other ML applications, i.e. works > very well in 90% of cases and fails dramatically in 10% of edge cases. Even >

Re: [Freetel-codec2] Codec2 vs Lyra?

2021-02-28 Thread David Rowe
Greg - LPCNet quantised to 3000 kbit/s would indeed be an interesting comparison; Jean-Marc and I have been quantising LPCNet at beneath 2000 bit/s, in my case to squeeze it through a HF Radio channel. I suspect the focus on 3000 bit/s is because that is the lowest rate that makes sense for

Re: [Freetel-codec2] Codec2 vs Lyra?

2021-02-28 Thread Adrian Musceac
While interesting and newsworthy, I'd assume from the start that this codec has the same advantages and pitfalls as other ML applications, i.e. works very well in 90% of cases and fails dramatically in 10% of edge cases. Even though the page specifies it is aimed at a completely different

Re: [Freetel-codec2] Codec2 vs Lyra?

2021-02-27 Thread Greg Maxwell
It's unfortunate that they didn't compare to LPCnet which would be the nearest comparable tech. Opus at 6kbps is really pushing it to the breaking point (at least with the current encoder). The results sound very impressive... though now I want to hear a 700bps version-- that's below the point

Re: [Freetel-codec2] Codec2 vs Lyra?

2021-02-27 Thread David Rowe
Hi Michael, Thanks for the post - very interesting.  Sure is an exciting time for speech coding. The speech quality of Lyra is vastly better than anything Codec 2 can offer at a similar bit rate (3000 bits/s).  Codec 2 would sound closer to the Speex 3 kbit/s samples on the Lyra page. Lyra

[Freetel-codec2] Codec2 vs Lyra?

2021-02-27 Thread mgraves mstvp . com
Hi, While I've watched from afar, this is my first message to this list. I was wondering how Codec 2 compares to this latest effort from Google; Lyra. https://ai.googleblog.com/2021/02/lyra-new-very-low-bitrate-codec-for.html?m=1 Michael Graves mgra...@mstvp.com o: