Multiformat 48 kbps Listening Test Poll

Topic: Multiformat 48 kbps Listening Test Poll (Read 19358 times) previous topic - next topic

0 Members and 1 Guest are viewing this topic.

Multiformat 48 kbps Listening Test Poll

Reply #25 – 2006-07-25 13:28:29

Quote from: haregoo on 2006-07-24 23:17:48

FYI, guruboolez presented 80 kbps personal listening test (summer 2005).

Thanks, I wasn't aware of that.

Quote

I doubt whether enough listener will be interested in uncommon bitrate test.

Quite a few people said in one of the earlier threads about the proposed 48 kbps test that they'd like to see higher bit rates tested though.

Anyway, I've seen quite a few people say that there's no point in having any more 128 kbps tests because virtually all of the codecs tested in the last one were virtually transparent at 128 kbps. And although I'm sure there will be future 128 kbps tests, it's true that they won't provide much new information, because all they'll show is how much closer codecs can get to the all-elusive score of 5.0.

So surely the way forward is to see how good performance has become at bit rates lower than 128 kbps?

Multiformat 48 kbps Listening Test Poll

Reply #26 – 2006-07-25 15:10:10

Nero's AAC encoder has been completely rewritten compared to the one that was tested in 2005.

Multiformat 48 kbps Listening Test Poll

Reply #27 – 2006-07-27 01:17:47

What about 112kbps then?

Multiformat 48 kbps Listening Test Poll

Reply #28 – 2006-07-27 04:47:10

Too near to 128. I want 96 or lower.

Multiformat 48 kbps Listening Test Poll

Reply #29 – 2006-07-27 09:59:03

What about a test of MP3 encoders?

There are some things that would be extremely interesting to find out, such as the quality of "fast" encoders like Helix vs. LAME...

Also for a lot of people MP3 is still very relevant.

Multiformat 48 kbps Listening Test Poll

Reply #30 – 2006-07-27 10:30:35

But if Nero's AAC encoder has been completely rewritten since this test:

80 kbps test results

don't you want to see how much it's improved by?

Multiformat 48 kbps Listening Test Poll

Reply #31 – 2006-07-27 11:06:14

We could run both if there are enough testers (of course, with a few months pause inbetween).

Multiformat 48 kbps Listening Test Poll

Reply #32 – 2006-07-27 12:32:27

Quote from: digitalradiotech on 2006-07-27 10:30:35

But if Nero's AAC encoder has been completely rewritten since this test:

80 kbps test results

don't you want to see how much it's improved by?

I do. But while I agree with this:

Quote

I also doubt Nero HE-AAC will see any big improvements in 6 months other than maybe multichannel updates and stuff like this.

I only agree for <= 64kbps and >128kbps. For the range of 64-128kbps LC AAC, I do think performance will improve noticeably in the near future. So I do want to know how much better it is but not necessarily at this exact point in time

80bkps is prolly a very good test of the encoder' efficiency, but it's not a very common bitrate.

At 128kbps MP3, there has been some evolution, there's the interesting case of the uber-fast Helix, and it's one of the most commonly used combinations. There might also not be that many contenders, which makes the test easier on the listeners and faster to organize & process.

You can alternatively test both at the same time and see if the modern formats at 80kbps can compete with MP3 at 128kbps

Multiformat 48 kbps Listening Test Poll

Reply #33 – 2006-07-27 12:53:43

Garf, did I say that? It sounds familiar.

Anyways, I was referring to the suggestiong to test 48 kbps now and in early 2007 again. To me, it makes no point since HE-AAC won't see improvements at 48 kbps and Vorbis either. I also don't think ATRAC3plus is going to be updated, so if we test now at 48 kbps and later when WMA 10 is out, three contenders will score the same. That's why I think we should all wait for WMA before running the 48 kbps test.

An 80 kbps test could be interesting because 128 was already transparent for most users and the differences between the tested encoders were non-significant. As you said Garf, we can also compare 80 kbps vs. MP3 at 128 kbps.

As for the MP3 test, you target 128 kbps, right? What encoders other than LAME and Helix do you have in mind? I guess we can omit Blade, but what about Xing and a FhG one?

BTW, this is getting off-topic. Maybe a moderator can either split the thread or rename it and delete the poll since it's useless.

Multiformat 48 kbps Listening Test Poll

Reply #34 – 2006-07-27 15:47:06

Quote from: Garf on 2006-07-27 12:32:27

Quote from: digitalradiotech on 2006-07-27 10:30:35
But if Nero's AAC encoder has been completely rewritten since this test:

80 kbps test results

don't you want to see how much it's improved by?

I do. But while I agree with this:

Quote
I also doubt Nero HE-AAC will see any big improvements in 6 months other than maybe multichannel updates and stuff like this.

I only agree for <= 64kbps and >128kbps. For the range of 64-128kbps LC AAC, I do think performance will improve noticeably in the near future. So I do want to know how much better it is but not necessarily at this exact point in time

Fair enough.

Out of interest, is intensity stereo already included in your encoder, or is it going to be added to a future release, because I remember Ivan mentioning it in a thread a few months ago?

Quote

80bkps is prolly a very good test of the encoder' efficiency, but it's not a very common bitrate.

True, but it might encourage some people to use 80 kbps instead of 48 kbps HE-AAC.

Quote

At 128kbps MP3, there has been some evolution, there's the interesting case of the uber-fast Helix, and it's one of the most commonly used combinations. There might also not be that many contenders, which makes the test easier on the listeners and faster to organize & process.

You can alternatively test both at the same time and see if the modern formats at 80kbps can compete with MP3 at 128kbps

Yes, I think that's a good idea.

Multiformat 48 kbps Listening Test Poll

Reply #35 – 2006-07-30 22:35:38

That's interesting how some people are arguing that a bitrate (80 kb/s) is an "uncommon bitrate". Why is this even relevant? It's not like all of the codecs we're testing are "common codecs" or that Hydrogen Audio represents the common, average computer user.

If the test shows that codecs can get close to transparent at that bitrate, then there will be a reason for these bitrates to become "common" for demanding users like us.

In my opinion, ogg Vorbis is nearly transparent at around 80 kb/s... the sweet spot. I'd like to see a more objective test to confirm that idea of mine, and also see how the other codecs available compare.

The point of these listening tests isn't to flog a dead horse, but to experiment with and demonstrate the potential of codecs and settings that users normally might not want to try. Perhaps 80 kb/s being uncommon is a perfect reason for it to be tested.

Multiformat 48 kbps Listening Test Poll

Reply #36 – 2006-07-31 00:03:48

so the 80 kbs test would basically be interesting for vorbis users and for who else? (i would think that 64 - 128 is pretty much empty ground?)

Multiformat 48 kbps Listening Test Poll

Reply #37 – 2006-07-31 08:47:17

Quote from: smok3 on 2006-07-31 00:03:48

so the 80 kbs test would basically be interesting for vorbis users and for who else? (i would think that 64 - 128 is pretty much empty ground?)

It would be interesting for anyone who wants to maximize the space on their portable (especialy flash memory-based) DAP for one! The only format that has a chance of comparing is AAC and it will be interesting to see how far it has come along since 2005. At 64 kbps and below, AAC+ is pretty much the undisputed champion but how many DAPs currently support this SBR-enabled format (with the exception of the occasional cell phone)?

Multiformat 48 kbps Listening Test Poll

Reply #38 – 2006-07-31 09:15:34

80kbps is perhaps providing the same quality as mp3@128kbps 5 years ago.
Then, mp3@128kbps was considered to be interesting, so I think that 80kbps now might also be interesting.

Multiformat 48 kbps Listening Test Poll

Reply #39 – 2006-07-31 11:05:34

I explained in my own test (mentionned above) why I decided to test encoders from 80 kbps to ~180 kbps and why I didn't start with 64 kbps. Last year I was unsatisfied by all coding formats at ~64 kbps: both vorbis and HE-AAC (the most advanced coders at low bitrate) were performing poorly to my ears and were therefore unusable according to my own tolerence to artefacts. I started my listening tests at 80 kbps because a preliminary experience clearly revealed me that this value was more interesting quality-wise (again: to my own ears). And indeed, aoTuV ended the test with results close to the high anchor I used (CBR 128 instead of V5, because I didn't really trust LAME VBR at this time), at least with non-classical samples.

If I had to perform again the same test with modern encoders I would imagine serious changes in the results. HE-AAC would perform much better than what the old and crap (@moderators: to my ears) Nero 6's codec was able to produce and I wouldn't be surprised if it ended the test as winner, close to the high anchor with both classical and non-classical samples. I also expect from LC-AAC much better performance at 80 kbps. Nero's encoder seems to have now an interesting quality at this bitrate.

80 kbps is definitely a bitrate I like over 64 kbps. What bother me is WMAPro. A new listening test -at 48 kbps as well as 192 kbps- without WMAPro looks now incomplete...

Multiformat 48 kbps Listening Test Poll

Reply #40 – 2006-07-31 21:22:30

Quote from: guruboolez on 2006-07-31 11:05:34

.... Last year I was unsatisfied by all coding formats at ~64 kbps: both vorbis and HE-AAC (the most advanced coders at low bitrate)

Nothing changed since then. Most advanced codecs like Vorbis, Nero HE-AAC and even new one wma10pro are far from transparency at 64 kbits. For me also 80-96 kbps test will be interesting. And of course performance of LC and HE.

Multiformat 48 kbps Listening Test Poll

Reply #41 – 2006-07-31 22:12:58

I'm not looking for transparency at this bitrate (blind tests performed and published here revealed me that 180 kbps aren't transparent either), only for acceptable/decent/usable quality. While I still unlike vorbis at 64 kbps in several cases, modern HE-AAC encodings appear more often acceptable for my own taste. It's just a rough and unsure conclusion based on fragmented experience and only a complete test would give me a better knowledge about it.
From my own point of view, I'd say that HE-AAC¹ have progressed much more than Vorbis in one year and may offer a better efficiency. But again, it's more a feeling than a robust knowledge

___
¹ Nero Digital encoder at least: I didn't test any other HE-AAC implementation last year. The Helix encoder had probably performed much better but 80 kbps encodings weren't possible [there was a big gap between 64 and 96 kbps]

Multiformat 48 kbps Listening Test Poll

Reply #42 – 2006-08-02 18:23:54

Regardless of bitrates or codecs, I wish we could just get on with SOMETHING! We seem to be wasting a lot of time bickering over trivialities.

Multiformat 48 kbps Listening Test Poll

Reply #43 – 2006-08-03 20:03:18

Hey guys, I am currently in Bucharest, Romania because my grandmother died and I am going to fly back to Germany next week. Sorry for the delay.

Multiformat 48 kbps Listening Test Poll

Reply #44 – 2006-08-04 03:52:57

Sorry to hear that Sebastian. I hope your trip goes well.

Multiformat 48 kbps Listening Test Poll

Reply #45 – 2006-08-04 17:59:56

My condolences too.

Notice

Poll

When should the test begin?