IPB

Welcome Guest ( Log In | Register )

 
Reply to this topicStart new topic
Getting ready to transcode library to AAC
Eli
post Jan 18 2013, 14:59
Post #1





Group: Members
Posts: 1056
Joined: 16-October 03
Member No.: 9337



I am getting ready to transcode my ~2TB flac library to AAC-LC in an MP4 container and want to be sure I get it right.

My intention is to encode at ~96kbps TVBR with QAAC. The goal is for portability and streaming, and this seems to be a good quality level that will give reasonable listening (I still have the lossless archive).

Is TVBR widely supported? I know many years ago when I used to encode to LAME MP3 VBR ipods had a hard time with it.



I need to pick the right tool for transcoding and have some specific tasks that I want to achieve and can hopefully achieve all of them in one transcode step.

- Scale down album art automatically. Probably to <50kb and 500x500. Is this still to big?
--- 90+% of my albums have embedded and folder.jpg, a minority have one or the other. If they differ, I would like to scale down from the higher resolution one.

- Identify HDCD tracks, and encode to 24b on the fly

- Identify pre-emph files and de-emph

- Apply EBU R128 ReplayGain ( I still need to calculate EBU R128 ReplayGain values for my flac files)

Is there a tool that can do all of these?

This post has been edited by Eli: Jan 18 2013, 15:18


--------------------
http://forum.dbpoweramp.com/showthread.php?t=21072
Go to the top of the page
+Quote Post
IgorC
post Jan 18 2013, 15:23
Post #2





Group: Members
Posts: 1553
Joined: 3-January 05
From: ARG/RUS
Member No.: 18803



CVBR prevents the situations where bitrate goes too low. CVBR is actually on par with TVBR (if not better).

The word "true" doesn't imply better VBR.

This post has been edited by IgorC: Jan 18 2013, 15:32
Go to the top of the page
+Quote Post
Eli
post Jan 18 2013, 15:44
Post #3





Group: Members
Posts: 1056
Joined: 16-October 03
Member No.: 9337



QUOTE (IgorC @ Jan 18 2013, 09:23) *
CVBR prevents the situations where bitrate goes too low. CVBR is actually on par with TVBR (if not better).

The word "true" doesn't imply better VBR.



I guess the idea is that TVBR can adjust to what is required and is more optimized. If the bitrate goes too low then TVBR has not been optimized in the encoder. I will look at CVBR more closely.


--------------------
http://forum.dbpoweramp.com/showthread.php?t=21072
Go to the top of the page
+Quote Post
robertcollier4
post Jan 18 2013, 18:23
Post #4





Group: Members
Posts: 33
Joined: 25-November 12
Member No.: 104754



Results of the public AAC listening test @ 96 kbps (July 2011)

Bootstrap Analysis
Nero CVBR TVBR FhG CT low_anchor
3.698 4.391 4.342 4.253 4.039 1.545

ANOVA Analysis
CVBR TVBR FhG CT Nero low_anch
4.39 4.34 4.25 4.04 3.70 1.55

CVBR scores a tiny tiny tiny bit above TVBR - but marginally (4.391 vs 4.342 for Bootstrap and 4.39 vs 4.34 for ANOVA).

In return for that tiny marginally higher score, the 'Bitrate table' shows that the mean bitrate of CVBR was 101 vs 94 for TVBR.

So - CVBR can give you possibly a marginally better quality - but at an average 7% increase in filesize.

Both TVBR and CVBR score tops in quality and significantly better than the rest of the competition - and in my opinion the 7% average filesize improvement overweighs the marginal quality gain given by CVBR over TVBR.

The comparison isn't really fair because equivalent filesizes aren't being compared - and if you are willing to go for the 7% bigger filesize - then you could rather just bump up the TVBR quality setting and probably get a better TVBR quality vs CVBR quality at the same filesize.

This post has been edited by robertcollier4: Jan 18 2013, 18:36
Go to the top of the page
+Quote Post
Soap
post Jan 18 2013, 18:36
Post #5





Group: Members
Posts: 1014
Joined: 19-November 06
Member No.: 37767



QUOTE (robertcollier4 @ Jan 18 2013, 12:23) *
CVBR scores a tiny tiny tiny bit above TVBR - but marginally (4.391 vs 4.342 for Bootstrap and 4.39 vs 4.34 for ANOVA).


The error bars overlap much more than the difference in their scores. A meaningful distinction between their relative qualities can not be drawn.

One's 7% larger. End of story.


--------------------
Creature of habit.
Go to the top of the page
+Quote Post
IgorC
post Jan 19 2013, 00:34
Post #6





Group: Members
Posts: 1553
Joined: 3-January 05
From: ARG/RUS
Member No.: 18803



QUOTE (robertcollier4 @ Jan 18 2013, 14:23) *
The comparison isn't really fair because equivalent filesizes aren't being compared - and if you are willing to go for the 7% bigger filesize - then you could rather just bump up the TVBR quality setting and probably get a better TVBR quality vs CVBR quality at the same filesize.

You're right.
Just one clarification. There was no way to match the bitrates between TVBR and CVBR.
TVBR can produce ~94 kbps or ~107-110 kbps but not ~100-101 kbps as CVBR 96 kbps. People had interest to test TVBR vs CVBR so we have decided to include the most close bitrate settings for both. Also testing TVBR at ~107 kbps to strongest competitor Fraunhofer AAC (new Winamp 5.62, VBR 3) ~100 kbps wouldn't be fair.

P.S. Even if both VBR modes would be at ~100 kbps still their scores will be practically the same. This personal test confirms it.

This post has been edited by IgorC: Jan 19 2013, 01:06
Go to the top of the page
+Quote Post
Dynamic
post Jan 19 2013, 14:04
Post #7





Group: Members
Posts: 803
Joined: 17-September 06
Member No.: 35307



To the OP,

Foobar2000, CUETools and dBpowerAmp all cover a lot of the aspects you require and copy over your tags happily.

The first part - about your AAC encoder choice and settings is answered above. I don't believe CUETools lets you use a commandline encoder, so might not be the choice here. I've often been pleasantly surprised that speech, mono and near-mono files and some vintage material often come out at greatly reduced bitrates using VBR (so TVBR might come into its own in such instances)

HDCD: I think CUETools will detect HDCD if you set that option, and foobar2000 will also detect it using foo_hdcd.

Foobar2000 can have the foo_hdcd encoder in the DSP chain of the Convert... function (separate from the playback DSP chain) and decode on the fly using its floating point high bit-depth pipeline to maintain maximum quality.

CUETools could pre-process any such albums to 20-bit in a 24-bit container, but I don't think it lets you access QAAC (only neroaacenc). It might be worth using CueTools to verify your lossless rips for accuracy anyway.

Might want to test them with one or two of your actual HDCD albums to ensure they apply it to all tracks on the same CD, because HDCD specifies a -6 dB volume adjustment to allow room for peak extension, so you might end up with inconsistent volume through the album.

Alternatively, give up on HDCD detection, as they sound pretty damned good undecoded. I should get round to ABXing the few I have at matched volume one day to see if it's worth it for me.

Pre-emphasis is something I've not experienced, but there are threads about it and lists of pre-emphasised releases, which you can look up to see if anything you own is among them, allowing you to test for detection. Definitely desirable to de-emphasise any such CDs.

Album Art: I've seen options regarding album art size in some tools I use (probably fb2k or CUETools), but haven't used them much.

There are various external tagging tools and the fb2k plugin foo_masstagger that might have options you want.

dBpowerAmp Converter is also pretty full of options developed for mass transcoding with very much the sort of requirements you have, and I think that includes tag checking against multiple sources (PerfectMeta) and album art choices too.

ReplayGain: I personally nearly always wish to play back at Album Gain volume even in non-RG-aware players (and I'm not too bothered whether it's the original or R-128 method that is used - they're much closer for Album Gain, a bit more varied with Track Gain), so I usually encode with foobar2000's Converter and choose Apply Album Gain in its ReplayGain settings before the audio is sent to the encoder. I then rescan after encoding to generate track gain (and all my Album Gain values are close to 0.00 dB). When I've ripped a load of lossless files I throw them into fb2k and scan as albums according to tags, which is very quick at scanning and tagging them with RG metadata and then ready for me to convert with gain applied. Whether you apply the gain before encoding or rely on your player all implementing RG from tags, it's still worth scanning after conversion.

I believe dBpowerAmp's DSP plugin system also lets you deal with ReplayGain when converting, though I'm unaware of whether it applies the EBU R-128 or original algorithm or gives you the option.

Overall, for my uses, I probably favour foobar2000 for this sort of task, but definitely like features of CueTools too and have found dBpowerAmp Converter's shell integration handy for speeding up some conversion tasks (e.g. FLAC to ALAC, or some quick Vorbis encodes to throw on my phone). I'd be quite happy to split the requirements among 2 or 3 programs.

Also, I'd probably make a text file in Notepad to paste in various settings for future reference (e.g. if I want to re-encode for another device, which might have, say, Opus support or a larger memory card), and after completion, I often also record the actual number of tracks encoded, duration, average bitrate etc. from the selection properties of fb2k. For example, I like to record my destination folder and file naming format, e.g.

%album artist%\%album% - %date%\%track% - %title% - %artist%

and also I record my encoder version and commandline switches (if specified as a custom encoder) and the bit-depth passed to it.

Best of luck, and please report your findings.

This post has been edited by Dynamic: Jan 19 2013, 14:12
Go to the top of the page
+Quote Post
nu774
post Jan 19 2013, 14:29
Post #8





Group: Developer
Posts: 522
Joined: 22-November 10
From: Japan
Member No.: 85902



QUOTE (Dynamic @ Jan 19 2013, 22:04) *
The first part - about your AAC encoder choice and settings is answered above. I don't believe CUETools lets you use a commandline encoder, so might not be the choice here.

CUETools lets you configure external CLI encoder from advanced setting dialog, and it has builtin presets for qaac TVBR mode by default.
Go to the top of the page
+Quote Post
Mach-X
post Jan 20 2013, 03:13
Post #9





Group: Members
Posts: 269
Joined: 29-July 12
From: Windsor, On, Ca
Member No.: 101859



I am questioning your goals here. If using ~96 kbps is your goal I assume you are looking to save space. If so, why the embedded album art, its a complete waste of space. Only windows media player/zune dont read folder.jpg. All other players like winamp and foobar do, as well as android and rockbox. (actually im not sure about itunes/ipod).


Go to the top of the page
+Quote Post
Eli
post Jan 21 2013, 15:56
Post #10





Group: Members
Posts: 1056
Joined: 16-October 03
Member No.: 9337



QUOTE (Mach-X @ Jan 19 2013, 21:13) *
I am questioning your goals here. If using ~96 kbps is your goal I assume you are looking to save space. If so, why the embedded album art, its a complete waste of space. Only windows media player/zune dont read folder.jpg. All other players like winamp and foobar do, as well as android and rockbox. (actually im not sure about itunes/ipod).


relying on folder.jpg requires you maintain the same folder structure, which may not be the case. Embedding is to make sure the art is there. I can probably scale down the size to 25kb max. However, I'm not sure that I can rely on having album art across multiple players if its not embedded. Am I wrong?


--------------------
http://forum.dbpoweramp.com/showthread.php?t=21072
Go to the top of the page
+Quote Post
spoon
post Jan 21 2013, 17:15
Post #11


dBpowerAMP developer


Group: Developer (Donating)
Posts: 2745
Joined: 24-March 02
Member No.: 1615



QUOTE
I believe dBpowerAmp's DSP plugin system also lets you deal with ReplayGain when converting, though I'm unaware of whether it applies the EBU R-128 or original algorithm or gives you the option.


It can write RBU R128, the R9 DSP effects (which are in beta test), and also all the other things on the list (except detect Pre-emphasis which is not detectable outside of a CD Ripper AFAIK)



--------------------
Spoon http://www.dbpoweramp.com
Go to the top of the page
+Quote Post
Eli
post Jan 21 2013, 18:03
Post #12





Group: Members
Posts: 1056
Joined: 16-October 03
Member No.: 9337



QUOTE (spoon @ Jan 21 2013, 11:15) *
QUOTE
I believe dBpowerAmp's DSP plugin system also lets you deal with ReplayGain when converting, though I'm unaware of whether it applies the EBU R-128 or original algorithm or gives you the option.


It can write RBU R128, the R9 DSP effects (which are in beta test), and also all the other things on the list (except detect Pre-emphasis which is not detectable outside of a CD Ripper AFAIK)


Spoon, I am well aware that dBpoweramp can do 95% of this. However, I don't think I can do it in one step with dBpoweramp.

In regards to pre-emphasis, most of mine are flagged in the meta-data thanks to to dBpoweramp flagging them. I'm sure some have slipped through in the past.

However, can I de-emph with dBpoweramp? I assume I would need to run SOX through the command line...

This post has been edited by Eli: Jan 21 2013, 18:04


--------------------
http://forum.dbpoweramp.com/showthread.php?t=21072
Go to the top of the page
+Quote Post
spoon
post Jan 21 2013, 22:18
Post #13


dBpowerAMP developer


Group: Developer (Donating)
Posts: 2745
Joined: 24-March 02
Member No.: 1615



I do not think you can easily de-emph with dBpoweramp and SOX together.


--------------------
Spoon http://www.dbpoweramp.com
Go to the top of the page
+Quote Post

Reply to this topicStart new topic
1 User(s) are reading this topic (1 Guests and 0 Anonymous Users)
0 Members:

 



RSS Lo-Fi Version Time is now: 20th August 2014 - 08:54