Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Worst psychoacoustic codec (Read 17454 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Worst psychoacoustic codec

What early psychoacoustic codec is generally considered to be the worst? I know MP1 was the first, but not the worst. K+K TAC gives me some pretty impressive artifacts even at a 1/4 compression ratio and there are codecs like Indeo Audio or Bink Audio which I haven't tried, but I suspect those might be "impressive" as well. AUPECg2 sounds funny as well, but very good for an university project done by 1 guy. Any famously bad codecs (not bad mp3 etc encoders, but formats that are pretty much universally bad)?

Worst psychoacoustic codec

Reply #1 –
Quote from: Neuron on
Any famously bad codecs (not bad mp3 etc encoders, but formats that are pretty much universally bad)?


How would you know that it is the codec and not the encoder, unless it has been so successful enough to have encoder development effort comparable to any of the "bigger" ones? You could if there were obvious errors by design (or maybe it wasn't even designed for musical fidelity), but otherwise?

Worst psychoacoustic codec

Reply #2 –
There were many codecs before MP1 that were not very good.  No idea if you can still find software for them.

Worst psychoacoustic codec

Reply #3 –
Quote from: saratoga on
There were many codecs before MP1 that were not very good.  No idea if you can still find software for them.


Know any names of them? I only know of ADPCM based codecs, but those are not psychoacoustic, and don't sound really bad either (the only ADPCM artifact is noise, none of that "wonderful" Blade/l3enc 0.99a warbly pre-echo flanging stuff). Were there any "official" audio codecs that used regular DCT instead of MDCT?

Worst psychoacoustic codec

Reply #4 –
Quote from: Neuron on
Quote from: saratoga on
There were many codecs before MP1 that were not very good.  No idea if you can still find software for them.


Know any names of them? I only know of ADPCM based codecs, but those are not psychoacoustic, and don't sound really bad either (the only ADPCM artifact is noise, none of that "wonderful" Blade/l3enc 0.99a warbly pre-echo flanging stuff). Were there any "official" audio codecs that used regular DCT instead of MDCT?


You'd have to dig through the literature.  Many of them probably don't even have names. 

Edit:  yes, there were FFT transform codecs before the MDCT.

Worst psychoacoustic codec

Reply #5 –
The first few generations of ATRAC encoders managed to introduce audible compression artefacts at 292kbps - that's pretty bad. Not the worst, and (as Porcus suggested) not fundamental to the format - just poor encoders.

Most earlier formats will sound terrible if you try to use them at the lower bitrates that are acceptable with modern lossy codecs. The vaguely successful ones all sounded OK at their target bitrates though.

I wonder if jj has something interesting tucked away from the early days?

Cheers,
David.

Worst psychoacoustic codec

Reply #6 –
I think 'worst' is hard to define, but...

Here in the Netherlands there has been some progression toward digital radio recently. They switched from DAB to DAB+, which for my favourite radio station meant switching from 192kbps MP2 to 64kbps HE-AAC v1. That 192kbps MP2 was definitely not transparent, while 64kbps HE-AAC v1 is most of the time. Not transparant at 192kbps is pretty bad IMO.

I don't know whether that's encoder or format dependent though. I don't think MP2 encoders have gotten as much attention as MP3-ones have. Looking at Great Britain and their use of DAB (so that's MP2) you'd think if there was still a lot of room for progression, they would invest quite some effort in it as switching to DAB+ like we did here won't happen anytime soon as most people overthere are used to DAB while it was still pretty new here.
Music: sounds arranged such that they construct feelings.

Worst psychoacoustic codec

Reply #7 –
Quote from: ktf on
[..]
I don't know whether that's encoder or format dependent though. I don't think MP2 encoders have gotten as much attention as MP3-ones have. Looking at Great Britain and their use of DAB (so that's MP2) you'd think if there was still a lot of room for progression, they would invest quite some effort in it as switching to DAB+ like we did here won't happen anytime soon as most people overthere are used to DAB while it was still pretty new here.

The UK's developments when it comes to digital radio are a good example of the difference between codec and encoder that Porcus refers to.

You'd think that there was no reason to do further develop and perfect MPEG 1, layer 2 (MP2) considering so many newer and better codecs have appeared since.

However, because of a large installed base of DAB-only radios (not capable of receiving the new DAB+ using AAC+) in the UK it will take some time for the UK to switch from MP2 to AAC+. The interesting result of this is that Coding Technologies noticed a market for a new MP2 encoder implementation. Over the last couple of years quite a few stations, including the BBC and Absolute, switched to a new encoder for a relatively ancient codec.
Every night with my star friends / We eat caviar and drink champagne
Sniffing in the VIP area / We talk about Frank Sinatra
Do you know Frank Sinatra? / He's dead

Worst psychoacoustic codec

Reply #8 –
MP2 radio is low quality because of transcoding, not the format itself. I tried MP2 and it is very decent for such a primitive codec. If your idea of terrible is "not transparent at 192 kbps" then you have not heard enough bad codecs. Search for ReallyRareWares here and copy the link into the Wayback machine, there are some pretty terrible codecs there. Hell, most old mp3 encoders are not transparent at 320 kbps, for example Blade is obvious almost always.

Download K+K TAC and please tell me, can anything be worse than that? It warbles, at 384 kbps. Not "I can hear the difference, barely" non transparent, but obvious, in your face artifacting.

I understand the encoder-codec difference, so I change my question to codecs that never got any decent encoders. Which of them was the worst? It is not about transparency, no encoder from that era was transparent, but which had the biggest and most annoying artifacts?

Worst psychoacoustic codec

Reply #9 –
Quote from: ktf on
192kbps MP2 to 64kbps HE-AAC v1.
In side by side tests, I prefer 192kbps mp2 to 64kbps HE-AAC. 192kbps mp2 can be transparent for some signals. Try toolame p2 discrete stereo 192kbps - it can even beat 192kbps CBR mp3 on some samples.

The audio part of DAB+ streams tends to run at a lower bitrate than that listed - some of that 64kbps is used for other things.

128kbps AAC - that's what DAB should have used. Did you know we now have entire DAB multiplexes in the UK with only mono audio broadcasting? Music stations, not just speech. 64kbps mono mp2. Obviously 64kbps HE-AAC would be a big improvement here.

Cheers,
David.

Worst psychoacoustic codec

Reply #10 –
Argh, DAB ... claimed to be an improvement 'cos it is digital, and in practice they use low bitrate, transcode too, and then the transition to DAB+ has the same incompatibilities as the transition from FM except now they want you to scrap expensive equipment ...

But I digress.

Worst psychoacoustic codec

Reply #11 –
So they went from decent (fm), to crap, and now to crap+? Why didn't they go with HD Radio? Piggybacks the analog signal, requires only 10% of the analog signal's transmission to reach the same area, cd quality, etc, etc.

Oh, and just to stay on topic, I'd say whatever codec Sirius/XM uses is pretty awful...

Worst psychoacoustic codec

Reply #12 –
Quote from: Mach-X on
Oh, and just to stay on topic, I'd say whatever codec Sirius/XM uses is pretty awful...

SiriusXM uses HE-AACv2. That is in itself a good codec of which high quality encoders and decoders are readily available.

However, it is up to the stations on Sirius to decide how much they care about sound quality. Most don't seem to care and opt for bitrates (often 24/32 kb/s) which aren't great for music. It's not the codec that's the problem here, it's how it's implemented.
Every night with my star friends / We eat caviar and drink champagne
Sniffing in the VIP area / We talk about Frank Sinatra
Do you know Frank Sinatra? / He's dead

Worst psychoacoustic codec

Reply #13 –
Quote from: Mach-X on
So they went from decent (fm), to crap, and now to crap+? Why didn't they go with HD Radio? Piggybacks the analog signal, requires only 10% of the analog signal's transmission to reach the same area, cd quality, etc, etc.
You face the same problems when they introduce extra stations though. They cut bitrates (hence quality) to cram more stations in. In some sense this is worse with HD Radio, because it's crammed into an existing, and often full, band.

The point of both systems is partly political. HD radio uses existing spectrum because the owners of that existing spectrum wanted to keep hold of it, and keep competition out. DAB uses new spectrum because governments (originally the UK government) wanted to see new broadcasters building viable businesses.

The technical benefit of DAB is that OFDM modulation works better (i.e. is more resistant to interference and multipath) over a wider bandwidth.

All broadcast systems are a collection of the best available/affordable/realisable technologies of their time. HD radio, coming later, got a better audio codec. It may have got slightly better error correction too - I haven't looked at it.

State of the art 2011 was DVB-T2 lite, which is about 8x more efficient than DAB. AFAIK no one is proposing to use it though. There are some new audio coding technologies from FhG which push things even further; if included they'll improve efficiency at low bitrates yet again. With a 4Mbps multiplex, and most stations using around 20kbps, the difficulty would be willing the space with FTA radio. It might get used for pay radio, but I wouldn't hold my breath.

Cheers,
David.

Worst psychoacoustic codec

Reply #14 –
Quote from: Neuron on
Download K+K TAC and please tell me, can anything be worse than that? It warbles, at 384 kbps. Not "I can hear the difference, barely" non transparent, but obvious, in your face artifacting.

Well, in that case, I don't think your question can really be answered. There's probably some unknown, obscure codec out there that performs worse.

I was just thinking "which codec has had quite some effort put into an encoder but still doesn't perform that well", and MP2 was the only one that qualified. As you changed the question from worst format to worst encoder, it's not relevant anymore.

Quote from: 2Bdecided on
In side by side tests, I prefer 192kbps mp2 to 64kbps HE-AAC.

I suppose the dutch public broadcaster used a bad encoder then.

Quote from: Mach-X on
So they went from decent (fm), to crap, and now to crap+? Why didn't they go with HD Radio?

It's not like crap+, HE-AAC @ 64kbit/s is pretty decent. They could have opted for a higher bitrate, but instead, they now broadcast twice the number of channels available in FM and there is still room for a lot more. They probably thought diversity was more important than having HiFi quality. I don't think HD Radio was really every an option here in the Netherlands, as DAB(+)'s single frequency network function solves quite a lot of problems with FM broadcasting. The FM spectrum is just too damn crowded here, population density is 14 times that of the USA
Music: sounds arranged such that they construct feelings.

Worst psychoacoustic codec

Reply #15 –
Why did rjamorim stop going on here anyways? I've read his old posts and he was pretty much an expert on this stuff.

Worst psychoacoustic codec

Reply #16 –
Why did rjamorim stop going on here anyways? I've read his old posts and he was pretty much an expert on this stuff.

And as other people said, MP2 can sound very good with a good encoder. Those 192 kbps mp2 streams were often reencoded up to 10 times, and a part of them carries non-audio data.

Worst psychoacoustic codec

Reply #17 –
Quote
It's not like crap+, HE-AAC @ 64kbit/s is pretty decent.

All our DVB-T2 radio stations are running at 64Kbps HE-AAC in South Africa and it sounds very annoying. I prefer FM radio.

Worst psychoacoustic codec

Reply #18 –
Quote from: Neuron on
What early psychoacoustic codec is generally considered to be the worst?

In 2000 the EBU did a listening test, and a codec called QDesign Music Codec 2 scored in the "poor" range even at 64 kbps stereo. The report of that test is here.

I don't know more about this codec than what Wikipedia mentions. Maybe the format itself allows for better quality, maybe not. I guess there's a reason, though, that we haven't heard of this codec lately.

Quote from: 2Bdecided on
The first few generations of ATRAC encoders managed to introduce audible compression artefacts at 292kbps - that's pretty bad. Not the worst, and (as Porcus suggested) not fundamental to the format - just poor encoders.

A colleague of mine once explained to me that the first-generation ATRAC format actually did have a design flaw (had to do with the cascaded filter banks) which made it sound so bad.

I agree that DAB+ should run at 96 kbps or more. The problem is not the codec (maybe the encoder is, possibly e.g. in South Africa?), the problem is that more than 10 kbps of the bit-rate are used for non-audio data.

Chris
If I don't reply to your reply, it means I agree with you.

Worst psychoacoustic codec

Reply #19 –
QDesign had very obvious artifacts even at 128 kbps, but it was meant for very low bitrates. It sounds a lot better at 20 kbps than mp3 of the same bitrate through. It was an agressive compression codec, but it doesn't really qualify as truly terrible as it did an OK job at extremely low bitrates in 1998 or so.

Worst psychoacoustic codec

Reply #20 –
What about AC1/AC2? Are there any encoders for those on the internet? AC3 sounds miserable at 128 kbps encoded by Aften, but completely OK at the typical DVD bitrates (although very critical listening might reveal pre-echo), but I suppose if AC3 sound bad at 128 kbps AC1 and AC2 must sound really bad. What about Intel Music Codec and Intel Indeo Audio as I mentioned before? I would make encodings myself, but Indeo doesn't install properly on my Windows 7 64-bit system and I don't have a Windows XP 32-bit machine at the moment. There are samples on MultimediaWiki, but only very low bitrate, around 20 kbps.

Worst psychoacoustic codec

Reply #21 –
Quote from: Neuron on
What early psychoacoustic codec is generally considered to be the worst? ... Any famously bad codecs (not bad mp3 etc encoders, but formats that are pretty much universally bad)?
VQF IST DEATH

Worst psychoacoustic codec

Reply #22 –
@Canar: From the user comments, he would find vqf as good codec. Remember that it was usually encoded at 96kbps for a "similar to mp3" quality.
Of course, it had its downsides, but it's not the type of bad codec he's looking for.

Worst psychoacoustic codec

Reply #23 –
Yeah. VQF sounds a lot better at 96 kbps than AC3 at 128 kbps and AC3 is not the worst codec ever, even if it is fairly weak at low bitrates.