HydrogenAudio

Lossy Audio Compression => Other Lossy Codecs => Topic started by: fabiorug on 2018-06-02 18:20:17

Title: Ghost codec news
Post by: fabiorug on 2018-06-02 18:20:17
When was the last news about the Ghost codec? https://people.xiph.org/~xiphmont/demo/ghost/demo.html
"First and foremost, Ghost is vaporware," Montgomery said. "At present it is merely a collection of ideas and some early-stage research. Eventually, it is intended to be a codec that improves upon and supersedes Vorbis in its current niche."
Title: Re: Ghost codec news
Post by: magicgoose on 2018-06-04 08:12:45
Interesting. I had a few ideas from these in my head for a while without reading about them (in particular, about trying to separate tonal signals and encoding them differently), it's kind of satisfying to see that someone else came up with them too.
But the page you linked seems to be last edited in 2011.
Title: Re: Ghost codec news
Post by: IgorC on 2018-09-08 17:35:13
Most part of open source codec developers are involved into development of functional video AV1 encoders like  https://github.com/xiph/rav1e/graphs/contributors
There's work to be done at least for several years (5+ to 10 years) to develop optimized AV1 encoder, VR, 3D and 360ยบ  AV1, AV2 (?) extensions.  And new compression of neural networks and so on.

As for now  AAC family and Opus are considered as more than enough good formats.

Title: Re: Ghost codec news
Post by: polemon on 2018-09-12 00:14:57
This has been discussed here not too long ago. The one area which kinda got overlooked it low-bandwith speech codecs, however the use cases are very different. In particular, those are telephony applications, often use on very band limited cases, such as digital voice radio. Quality is a secondary concern there, the primary being legibility.
Title: Re: Ghost codec news
Post by: foxyshadis on 2018-12-04 10:59:53
When was the last news about the Ghost codec? https://people.xiph.org/~xiphmont/demo/ghost/demo.html
"First and foremost, Ghost is vaporware," Montgomery said. "At present it is merely a collection of ideas and some early-stage research. Eventually, it is intended to be a codec that improves upon and supersedes Vorbis in its current niche."
If you look at the basic idea of Ghost, you'll see something interesting: The primary experimental features are tone analysis & removal, wavelet filters, and working in the time domain. Tone analysis became the basis for Opus, integrating that capability into CELT instead of leaving it for a future experimental codec. The other pieces ended up not really panning out.

So that's the status: Ghost is part of Opus now. Any future work will be a different name and ideas entirely.