Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Opus 1.5 released: Opus gets a machine learning upgrade

opus-codec.org

Post reply on HN
Print view View original

91–100 of 151 posts

Re: Opus 1.5 released: Opus gets a machine learning upgrade

#91

Post by jmvalin »

aredox wrote: post #82

>That's why most codecs have packet loss concealment (PLC) that can fill in for missing packets with plausible audio that just extrapolates what was being said and avoids leaving a hole in the audio ...How far can ML PLC "hallucinate" audio? A sound , a syllable, a whole word, half a sentence? Can I trust anymore what I hear?

What the PLC does is (vaguely) equivalent to momentarily freezing the image rather than showing a blank screen when packets are lost. If you're in the middle of a vowel, it'll continue the vowel (trying to follow the right energy) for about 100 ms before fading out. It's explicitly designed not to make up anything you didn't say -- for obvious reasons.

Re: Opus 1.5 released: Opus gets a machine learning upgrade

#92

Post by adgjlsfhk1 »

viraptor wrote: post #90
WithinReason wrote: post #89

Earlier quoted context omitted.

You can more accurately invert the quantisation step

You can't do it more accurately. You can make up expected details which aren't encoded in the file. But that's explicitly less accurate.

If the encoders know what model the decoders will be running, they can improve accuracy. You could pretty easily make a codec that doesn't encode high resolution detail if the decoder NN will interpolate it correctly.

Re: Opus 1.5 released: Opus gets a machine learning upgrade

#93

Post by a_wild_dandan »

CharlesW wrote: post #29
a_wild_dandan wrote: post #25

Earlier quoted context omitted.

To borrow from Joscha Bach: if you like the output, it's called creativity. If you don't, it's called a hallucination.

I love that, what's it from? (My Google-fu failed.) Unexpected responses are often a joy when using AI in a creative context. https://www.cell.com/trends/neurosciences/abstract/S0166-223...

It was from one of his podcast appearances. Which doesn't narrow it down much, unfortunately. Most likely options:

https://www.youtube.com/watch?v=LgwjcqhkOA4

https://www.youtube.com/watch?v=sIKbp3KcS8A

https://www.youtube.com/watch?v=CcQMYNi9a2w

Re: Opus 1.5 released: Opus gets a machine learning upgrade

#95

Post by anthk »

kristopolous wrote: post #85
Dwedit wrote: post #77

I just want to mention that getting such good speech quality at 9kbps by using NoLACE is absolutely insane.

I wanted to see what it would sound like in comparison to a really early streaming audio codec, realaudio 1.0 $ ffmpeg -i female_ref.wav - acodec real_144 female_ref.ra And if you can't support that I put it back to wav and posted it: http://9ol.es/female_ref-ra.wav This was seen as "14.4" audio, for 14.4kb/s dialup in the mid-90s. The quality increase over those nearly 30 years for what you can get out of what's act…

I used to listen opus avant agarde music from https://dir.xiph.org at 16kb/s under a 2G connection and it was usable once mplayer/mpv cached back the stream for nearly a minute.

Re: Opus 1.5 released: Opus gets a machine learning upgrade

#96

Post by Dylan16807 »

a_wild_dandan wrote: post #25
CharlesW wrote: post #5

Earlier quoted context omitted.

Isn't packet loss concealment (PLC) a form of hallucination? Not saying it's bad, just that it's still Making Shit Up™ in a statistically-credible way.

To borrow from Joscha Bach: if you like the output, it's called creativity. If you don't, it's called a hallucination.

Doesn't the context affect things much more than whether you like the particular results?

Either way, "creativity" in the playback of my voice call is just as bad.

Re: Opus 1.5 released: Opus gets a machine learning upgrade

#97

Post by shwaj »

thomastjeffery wrote: post #58
radarsat1 wrote: post #36

Earlier quoted context omitted.

This is an important question. However, I'd like to point out that similar biases can easily exist for non-ML, hand-tuned algorithms. Even in the latter case test sets and often even "training" and "validation" sets are used for finding good parameters. Any of these can be a source of bias, as can the ears of evaluators making these decisions. It's true that bias questions often come up in ML context because fundamen…

Usually regular algorithms aren't generating data that pretends to be raw data. That's the significant difference here.

Can you precisely define what you mean by "generating" and "pretends", in such a way that this neural network does both these things, but a conventional modern audio codec doesn't?

"Pretends" is a problematic choice of words, because it anthropomorphizes the algorithm. It would be more accurate and less misleading to replace "pretends to be" with "approximates". But then it wouldn't serve your goal of (seeming to) establish a categorical difference between this approach and "regular algorithms", because that's what a regular algorithm does too.

I apologize, because the above might sound rude. It's not intended to be.

Re: Opus 1.5 released: Opus gets a machine learning upgrade

#98

Post by nextaccountic »

unixhero wrote: post #47
luplex wrote: post #27

I wonder: did they address common ML ethics questions? Specifically: Are the ML algorithms better/worse on male than on female speech? How about different languages or dialects? Are they specifically tuned for speech at all, or do they also work well for music or birdsong? That said, the examples are impressive and I can't wait for this level of understandability to become standard in my calls.

Why is the ethics question important? It is a new feature for an audio codec, not a new material to teach in your kids curriculum.

Because this gets deployed in real world, affecting real people. Ethics don't exist only in kids curriculum.

Re: Opus 1.5 released: Opus gets a machine learning upgrade

#100

Post by shwaj »

viraptor wrote: post #86
cma wrote: post #76

Earlier quoted context omitted.

What if it is a pareto improvement: better improvement for some dialects but no worse than the earlier version for anyone. Should it be shelved or tuned down so all improvement for each dialect see gains by an exactly equal percentage?

Here's a question that should have the same/similar answer: Increasingly some part of the job interviews is being handled over the internet. All other things being equal, people are likely to have a more positive response to candidates with more pleasant voice. So if new ML-enhanced codecs become more common, we may find that some group X has a just slightly worse quality score than others. Over enough samples that w…

I don't think it's a given that we shouldn't keep using that codec. For example, maybe the improvement is due to an open source hacker working in their spare time to make the world a better place. Do we tell them their contribution isn't welcome until it meets the community's benchmark for equity?

Your same argument can also be used to degrade the performance for all other groups, so that group X isn't unfairly disadvantaged. Or, it can even be used to argue that the performance for other groups should be degraded to be even worse than group X, to compensate for other factors that disadvantage group X.

This is argumentum ad absurdum, but it goes to show that the issue isn't as black and white as you seem to think it is.

Post reply on HN