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: ~128 kbps Private Listening Test/ Multi Format (Read 9079 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

~128 kbps Private Listening Test/ Multi Format

Summary

ABC/HR for Java 0.53b listening tests with 6 VBR (+ 1 CBR) Codecs within a ~128 kbps range. 12 samples - 6 Musical Genres. The goal is to determine what codec achieves better quality with a compression range of ~128 kbps

Codecs

MP3: LAME 3.98.2  through Audacity 1.3.11 Beta
  • V5 with VBR
  • 110 - 150 Kbbps
  • Channe Mode: Joint Stereo
  • Variable Speed: Standard

Ogg Vorbis: aoTuV beta 5.7 though OggdropXPd 1.9.0
  • -quality 4.00 with VBR/SBR
  • Nominal Bitrate of 128 kbps

AAC: Nero 1.5.1 through foobar2000 1.0
  • Quality 0.41
  • ~130 kbps

QuickTime/Apple (Encoder) through iTunes 9.0.3.15
  • Stereo Bit Rate: 128 kbps with VBR
  • Sample Rate: Auto
  • Channel Rate: Auto

Windows Media Audio Standard 9
  • VBR
  • Bit Rate Range of 85 - 145 kbps

Windows Media Audio Professional 10
  • CBR
  • 128 kbps

Musepack SV8 through foobar2000 1.0
  • Quality Setting: Q4 (~128 kbps)


Samples

6 Genres:
Classical, Jazz, Rock, Pop, Hip-Hop/Rap, Country
10 seconds per sample

Classical:
Drew Tretick: A Summer Serenade
  • Track 8 - O mio babbino caro
  • Track 9 - Flight of the Bumblebee


Jazz:
20 Best of Jazz
  • Track 19 - What is Thing Called Love? by Charles Mingus
  • Track 7 - People by Art Farmer


Country:
NOW That's What I Call Country
  • Track 2 - Winner At A Losing Game by Rascal Flatts
  • Track 16 - You're Gonna Miss This by Trace Adkins


Rock:
  • Album - Led Zeppelin IV; Artist - Led Zeppelin; Track 1 - Black Dog
  • Album - Pablo Honey; Artist - Radiohead; Track 3 - How Do You?


Pop:
  • Album & Artist - The Good The Bad The Queen; Track 3 - Northern Whale
  • Album - Rock Steady; Artist - No Doubt; Track 4 - Making Out


Hip-Hop/Rap:
  • Album & Artist - Gorillaz; Track 5 - Clint Eastwood
  • Album - All World; Artist - LL Cool J; Track 2 - Rock the Bells




Thoughts? Questions?

~128 kbps Private Listening Test/ Multi Format

Reply #1
Once again why don't you just use lame mp3 3.98.3,Nero AAC 1.5.4 (although it's bugfix related) and quicktime true vbr with nao app,all through foobar2000? It's easier and you save a lot of time

~128 kbps Private Listening Test/ Multi Format

Reply #2
For LAME, the "standard" variable bitrate mode is the old mode. Use the one labeled fast (which is the default in lame 3.98 and up).



~128 kbps Private Listening Test/ Multi Format

Reply #4
*New Samples, encoders and frontends:

Codecs

MP3: LAME 3.98.3 through Foobar2000 v1.0

    * V5 with VBR
    * ~130 Kbps


Ogg Vorbis: aoTuV beta 5.7 though OggdropXPd 1.9.0

    * -quality 4.00 with VBR/SBR
    * Nominal Bitrate of 128 kbps

The reason why I'm using oggdropXPd for Ogg Vorbis is because I want to use aoTuV b5.7. I don't really know how to adjust the aoTuV libraries to work perfectly with foobar2000. Besides, oggdropXPd is pretty simple and intuitive to use.


AAC: Nero 1.5.1 through foobar2000 1.0

    * Quality 0.41
    * ~130 kbps


QuickTime/Apple (Encoder) through iTunes 9.0.3.15

  1. Stereo Bit Rate: 128 kbps with VBR
  2. Sample Rate: Auto
  3. Channel Rate: Auto


Windows Media Audio Standard 9

    * VBR
    * Bit Rate Range of 85 - 145 kbps


Windows Media Audio Professional 10

    * CBR
    * 128 kbps

For both WMA formats, I burned the original WAV samples on to a disc, and then ripped them with WMP for both WMA Std 9 & Pro 10

Musepack SV8 through foobar2000 1.0

    * Quality Setting: Q4 (~128 kbps)



Samples

6 Genres (+ 1 Low Anchor):
Classical, Jazz, Rock, Pop, Hip-Hop/Rap, Country
9 - 11 seconds per sample

Classical:
Drew Tretick: A Summer Serenade

    * Track 2 - The Prayer (After Foster)
    * Track 6 - Nocturne Op 9 No 2

Jazz:
20 Best of Jazz - Various Artists

    * Track 7 - People by Art Farmer

New Concepts of Artistry - Stan Kenton

    *Track 7 - Young Blood



Country:
NOW That's What I Call Country - Various Artists

    * Track 1 - All-American Girl by Carrie Underwood
    * Track 3 - Watching Airplanes by Gary Allen



Rock:

    * Album - Led Zeppelin IV; Artist - Led Zeppelin; Track 2 - Rock and Roll
    * Album - Silver Side Up; Artist - Nickleback; Track 8 - Where Do I Hide?



Pop:

    * Album & Artist - The Good The Bad The Queen; Track 5 - Herculean
    * Album - Rock Steady; Artist - No Doubt; Track 411 - Platinum Blonde



Hip-Hop/Rap:

    * Album & Artist - Gorillaz; Track 1 - Re-Hash
    * Album - All World; Artist - LL Cool J; Track 1 - I Can't Live Without My Radio

Low Anchor:

Unknown Sample (Need to decide on a codec and quality/bit rate setting) Here are some options:

* Ogg Vorbis with aoTuV b5.7 (-q 1.00 [~80kbps]

* LAME 3.98.3 MP3 (V9 [65])

* WMA Pro 10 @45 kbps

I changed the samples because ABC/HR for Java 0.53a only works with certain samples. It's a really annoying issue, and it took me quite a while to find samples that worked with the program. You can find more about the issue on another topic: http://www.hydrogenaudio.org/forums/index....showtopic=79468



~128 kbps Private Listening Test/ Multi Format

Reply #5
Getting aotuv b5.7 to work in foobar is very easy you just have to download the modified oggenc from aotuv site and change the name to that which foobar needs and that's it, it's going to use aotuv b5.7 perfectly

~128 kbps Private Listening Test/ Multi Format

Reply #6
Quote
Getting aotuv b5.7 to work in foobar is very easy you just have to download the modified oggenc from aotuv site and change the name to that which foobar needs and that's it, it's going to use aotuv b5.7 perfectly


I can understand where your going, but I've already prepared the samples, and going through the process again would mean I would have to re analyze what files work with ABC/HR 0.53a. It's a really annoying issue that I faced, and it took me a couple days to work around it. If the bug was somehow fixed, or if I found out what's wrong with the program, I would seriously go back and re - prepare the samples. Trust me, oggdropxpd is just as fast and efficient as foobar, so its not a big issue



~128 kbps Private Listening Test/ Multi Format

Reply #7
Geez... felix26591, if you reply in a thread and your reply is related to the post right above, don't quote the whole post especially if the post is 1km long. Quite annoying when surfing HA via netbook (600 px vertical resolution - toolbars - window decoration, etc.) or iPhone.

Anyways, back to topic. Like I mentioned via PM already, I think 45 kbps WMA Pro is too low as low anchor. I would choose 64 kbps because Microsoft always claimed that WMA at 64 kbps is as good as / beats MP3 at 128 kbps. While this was never true for WMA Standard, with WMA Professional they are one step closer to truth.

~128 kbps Private Listening Test/ Multi Format

Reply #8
Geez... felix26591, if you reply in a thread and your reply is related to the post right above, don't quote the whole post especially if the post is 1km long. Quite annoying when surfing HA via netbook (600 px vertical resolution - toolbars - window decoration, etc.) or iPhone.

Anyways, back to topic. Like I mentioned via PM already, I think 45 kbps WMA Pro is too low as low anchor. I would choose 64 kbps because Microsoft always claimed that WMA at 64 kbps is as good as / beats MP3 at 128 kbps. While this was never true for WMA Standard, with WMA Professional they are one step closer to truth.


I guess so. I've been considering using the Ogg Vorbis option I mentioned, because I've noticed that LAME 3.98.3 with both foobar and audacity always trips up to around 256 kbps with some samples. Then again, I can't really prove how often each encoder/frontend uses high compression rates, since I'm using VBR for the MP3 samples


 

~128 kbps Private Listening Test/ Multi Format

Reply #10
MP3 is also quite low with 65 kbps. Vorbis is fine IMO.


Vorbis it is then. For the moment anyways.

I know this may seem as a redundant question, but the tests participants don't need to know that there will be a low anchor, right? I've been sticking to Roberto Amorim's listening test document as a guide, but it doesn't mention anything about anchors....


~128 kbps Private Listening Test/ Multi Format

Reply #12
No idea to be honest. I think MUSHRA for example states that both anchor and reference are hidden, but I didn't use MUSHRA in my tests for example.


Oh, ok. Is Roberto ever online anymore? I've had some questions that I wanted to ask him, but his profile says that he's been offline for quite a while. If anything, I won't mention anything about the low anchor. From what iv'e heard, all the codecs have seemed quite transparent for the most part, so using something like -q 1.00 for Ogg Vorbis would seem pretty identifiable amongst the other codecs.

~128 kbps Private Listening Test/ Multi Format

Reply #13
Musepack SV8 through foobar2000 1.0

    * Quality Setting: Q4 (~128 kbps)


I've tested random 42 files (about 2 GB) of jazz, art rock, hard & heavy with Q4 (SV8) and I've got quite different bitrates, between 104 and 160 kbps.
So, you should play a bit with Q parameter if you want to compare the codecs at about 128 kbps.


~128 kbps Private Listening Test/ Multi Format

Reply #15
wow, the way that other thread was going I thought this was never going to happen

~128 kbps Private Listening Test/ Multi Format

Reply #16
By the way, would any of you like to participate in the tests? I already have some people that have taken the tests and its nearly done, but I think it would be a good idea to stretch the test results. I'm finishing up to,tomorrow, so  PM me if your interested.