Welcome to Doom9's Forum, THE in-place to be for everyone interested in DVD conversion.

Before you start posting please read the forum rules. By posting to this forum you agree to abide by the rules.

 

Go Back   Doom9's Forum > General > Audio encoding

Reply
 
Thread Tools Search this Thread Display Modes
Old 18th November 2007, 19:29   #1  |  Link
fenomeno83
Registered User
 
Join Date: Aug 2006
Posts: 330
switch from aac lc to he(v1 and 2)

Which are the range to switch the aac mode?
For example I found these (for stereo mode):
for bitrate <96 change from aac lc to aac he v1
for bitrate <48 change from aac he v1 to aac he v2

In another forum I found different ranges:
<85 switch from aac lc to aac he v1
<40 switch from aac he v1 to aac he v2

Which are the real range (for example in nero)?

And in every ranges,which sampling in used?

And which are bitrates to get a good sound?
For example:
for encode a original wav song which is the minimum good quality bitrate and tha maximum excellent quality bitrate(if the increase quality of a bigger bitrate isn't significant,I prefer that you write me the lesser bitrate..for example if you get an excellent quality al 128 kbit/s and at 160 kbit/s the sound is better only of 1%,I prefer that you write 128 and not 160)

And for film (from ac3 to aac stereo)which is the minimum good quality bitrate and the maximum excellent quality bitrate?
thanks

Last edited by fenomeno83; 18th November 2007 at 19:38.
fenomeno83 is offline   Reply With Quote
Old 18th November 2007, 20:24   #2  |  Link
tebasuna51
Moderator
 
tebasuna51's Avatar
 
Join Date: Feb 2005
Location: Spain
Posts: 5,248
Don't exist exact values because don't exist exact methods to measure the 'quality'. There are differents kinds of music, players, speakers, ears, ...

The differences between 40-48 or 85-96 are inside the personal choice. Maybe is better let the encoder select the bitrate for the exact sample with a quality parameter instead with a fix bitrate.

We can have excellent quality with 96 for an audio track with only speech but very bad for an audio track from a concert. Nothing is fix all is relative.

Last edited by tebasuna51; 10th February 2008 at 03:49.
tebasuna51 is offline   Reply With Quote
Old 18th November 2007, 20:49   #3  |  Link
fenomeno83
Registered User
 
Join Date: Aug 2006
Posts: 330
and about the switching bitrate used by nero e its sampling?
fenomeno83 is offline   Reply With Quote
Old 19th November 2007, 20:56   #4  |  Link
montython
Registered User
 
Join Date: Oct 2007
Posts: 22
I raised up a similar question in another topic. I know very little on the subject of audio coding, the following is what I have gathered after a rough survey:

2 diagrams to give a general idea of the operating ranges:


Source: Coding Technologies, EBU Technical Review (2006)


Source: Convention Paper 6199, Audio Engineering Society

These diagrams suggest that HE-AAC v1 (aac+sbr) is optimal below 96 kbps level. Also, PS (parametric stereo) option is relevant below 48 kbps. It should be noted that these numbers are for the sampling rates of 44,100 and 48,000.

What about different sampling rates? This is how command-line Nero AAC encoder (v1.1.34.2) operates:



As far as I can understand, these are just predefined numbers which do not depend on the material to be encoded. It seems that these predefined numbers are more or less appropriate for sampling rates around 44,100. The automatic choices of the codec for different sampling rates doesn't seem to be plausible and/or consistent. At 50 kbps level, it doesn't make much sense to turn on SBR for mono-11,025 sampling rate input while using LC-AAC profile for mono-48,000 sampling rate input. Possibly, the internals of the codec was not designed having these lower sampling rates in mind.

Is there a rule of thumb that could apply to all sampling rate levels?
Presence of more than one channels will complicate the issue, but the following generalizations might be handy:
  • Plain aac codec is likely to produce "near-transparency" results with compression ratios in the range 1:10 - 1:15.
  • At 32kbps bitrate level, even an mp3 codec can produce good results without noticable artifacts for mono input at 22,050 sampling rate. (which translates to approximately 1:10 compression ratio)
Based on this, a rough guide for HE-AAC might be the following:


The lower and upper limits at each sampling rate are the bitrate levels corresponding to 1:10 and 1:15 compression ratios. Switching to HE-AAC v1 and v2 makes sense below these ranges with a margin. Obviously, the actual required bitrate will depend on the content. These numbers are here just to give an idea.


p.s.
16-bit resolution is assumed everywhere.
montython is offline   Reply With Quote
Old 19th November 2007, 23:32   #5  |  Link
fenomeno83
Registered User
 
Join Date: Aug 2006
Posts: 330
thankyou very much...
I understand that between 96 and 140 kbit/s I get a perfect(or similar to perfect)quality..for these reason I can get the perfect qualit only in the aac lc profile...

For good quality I can choose lesser bitrates as 48(maybe is also good for film)64 or 80 but use HE profile (version 1)
HE vers 2 is very good for streaming application for bitrate as 24 or 32 or 40 kb/s but HE vers 2 isn't used for encode music or film

For me the best solution is:
use aac lc with bitrate beetween 96 and 140 (128 is suggested)to encode music or live concert dvd using 44100 or 48000 sampling (44100 in suggested),stereo
use aac he vers 1 to encode film with bitrate between 48 and 80 kb/s (64 is suggested) using 32000 sampling(44100 isn't necessary ?),stereo
use aac he vers 2 only for streaming application with bitrate between 16 and 40 kb/s using 22050 sampling..but in this case is better use mono

It's ALL true?If no,how do you change my paramethers?

Last edited by fenomeno83; 19th November 2007 at 23:43.
fenomeno83 is offline   Reply With Quote
Old 21st November 2007, 07:32   #6  |  Link
montython
Registered User
 
Join Date: Oct 2007
Posts: 22
This is the spectrogram display of a 16-bit, 44100 Hz sampling rate input:

In this case, downsampling this file into 22050 Hz sampling rate would not suffer from noticable quality loss.

Consider this one:

Nyquist–Shannon sampling theorem suggests that we need at least a sampling rate of 44100 Hz. Otherwise, it will not be possible to cover the whole frequency range of the input.

So, the required sampling rate depends on the input.
Quote:
Originally Posted by fenomeno83 View Post
use aac he vers 1 to encode film with bitrate between 48 and 80 kb/s (64 is suggested) using 32000 sampling(44100 isn't necessary ?),stereo
Assume that we have a stereo input similar to the one in the second figure which requires a sampling rate of 44100. Our target bitrate is 64 kbit/s.
The starting point should be the following:
At this bitrate level, can the aac encoder produce good results for 16-bit, stereo, 22050 Hz input? The reference table in my previous post suggests that the bitrate range of 45-70 is likely to produce very good results for such input.

So, the underlying coder will be able to successfully transmit the lower part of the spectrum. On the other hand, the higher frequencies will be reconstructed by the SBR decoder using the information included in the bitstream.

In conclusion, at 64 kbit/s it makes sense to stay at 44100 Hz and rely on SBR.

Quote:
Originally Posted by fenomeno83 View Post
use aac he vers 2 only for streaming application with bitrate between 16 and 40 kb/s using 22050 sampling..but in this case is better use mono
Again referring to the table in my previous post, a bitrate level in the range 45-70 is likely to produce very good results for 16-bit, stereo, 22050 Hz input. If our target bitrate is only slightly below this (40 kbit/s for example), parametric stereo is not really needed. AAC+SBR will be the optimal choice.

So, when do we need HE-AAC v2 (aac+sbr+ps)?

Assume that we have a 16-bit, 44100 Hz, stereo input. Our target bitrate is 32 kbit/s. At this bitrate level, can the aac encoder produce good results for 16-bit, stereo, 22050 Hz (half the original sampling rate) input? The table suggests that we need 45-70 kbit/s, so the answer is no. From this information, we might expect that HE-AAC v1 will not produce satisfactory results.

The next question is the following:
At bitrate levels slightly below (some datarate should be reserved for guidance sbr and ps info) our target bitrate (32 kbps), can the aac encoder produce good results for 16-bit, mono, 22050 Hz input?

The answer is yes. In this case, it makes sense to use HE-AAC v2.


p.s.
What I have written here are somewhat cheap generalizations relying on a brief reading. Actually, I don't have any background on this subject. I suggest that anyone reading my post should consider this.

Last edited by montython; 21st November 2007 at 07:35.
montython is offline   Reply With Quote
Old 22nd November 2007, 10:36   #7  |  Link
fenomeno83
Registered User
 
Join Date: Aug 2006
Posts: 330
THANK YOU VERY MUCH!

But now I didn't understand very well the ranges
Assuming that bitrate avaliable are
16-24-32-40-48-56-64-80-96-112-128-160-192-256-320-448 kb/s

If I must encode between 16 and 32 use aac he v2,22050hz,mono.stereo is impossibile to use in aac he v2?

If I must encode between 40 and 80 use aac he v1,22050,stereo(you tell 70,but is better assuming 64 or 80 as the maximum?)...22050 is good for all bitrate in the range(40-48-64-80(?) )or for someone is necessary use 32000 or 44100 hz?

If I must encode between 96(I don't now if is better assuming 80kb/s in aac he v1 or aac lc) and 160 is better to use aac lc (I will get a quality perfect!),44100(or 48000?),stereo

Bitrates 192-256-320-448 aren't necessary

Thank you
fenomeno83 is offline   Reply With Quote
Old 22nd November 2007, 19:16   #8  |  Link
montython
Registered User
 
Join Date: Oct 2007
Posts: 22
Quote:
Originally Posted by fenomeno83 View Post
If I must encode between 16 and 32 use aac he v2,22050hz,mono.stereo is impossibile to use in aac he v2?
No, no, you misunderstood me. HE-AAC v2 (aac+sbr+ps) makes sense only for stereo input anyway. Parametric stereo (ps) for mono input is useless.

What I say is this:
Quote:
Assume that we have a 16-bit, 44100 Hz, stereo input. Our target bitrate is 32 kbit/s.
(...)
At bitrate levels slightly below (some datarate should be reserved for guidance sbr and ps info) our target bitrate (32 kbps), can the aac encoder produce good results for 16-bit, mono, 22050 Hz input?

The answer is yes. In this case, it makes sense to use HE-AAC v2.
To make it more clear:
If plain aac can produce good results for 16-bit, mono, 22050 Hz input at 32 kbps, then it means that HE-AAC v2 can produce satisfactory results for 16-bit, stereo, 44100 Hz input at the same bitrate. So, at 32 kbit/s, use HE-AAC v2 for 16-bit, stereo, 44100 Hz input.

This is the logic: AAC part encodes the lower frequencies (at half the original sampling rate) of a mono stream. SBR part handles the higher frequencies, so that we end up with an output at 44100. Finally, PS part regenerates a stereo output, so that we end up with stereo, 44100 Hz output.



Quote:
Originally Posted by fenomeno83 View Post
If I must encode between 40 and 80 use aac he v1,22050,stereo(you tell 70,but is better assuming 64 or 80 as the maximum?)...22050 is good for all bitrate in the range(40-48-64-80(?) )or for someone is necessary use 32000 or 44100 hz?
What I say is this:
At a given bitrate level: If the plain AAC can produce good results for 16-bit, stereo, 22050 Hz input; then AAC+SBR can produce good results for stereo, 44100 Hz input. (Follow the logic given in the example above)

So, at 64 kbit/s, there is no need to downsample your input, or to convert to mono. Just keep it stereo at 44100 Hz, and HE-AAC v1 will handle it.


Quote:
Originally Posted by fenomeno83 View Post
If I must encode between 96(I don't now if is better assuming 80kb/s in aac he v1 or aac lc) and 160 is better to use aac lc (I will get a quality perfect!),44100(or 48000?),stereo
If the target bitrate is above 96 kbit/s, prefer LC-AAC.
80 kbit/s with HE-AAC v1 vs 96 kbit/s with LC-AAC is a personal choice, I cannot comment on that.

About the samplig rate: 44100 or 48000?
Keep the original sampling rate unless you have a good reason to resample and you know what you are doing.

Quote:
Originally Posted by fenomeno83 View Post
Bitrates 192-256-320-448 aren't necessary
192 kbit/s is usually considered to be a bitrate level which ensures transparency. (not exactly the same with the original, but the differences are inaudible to human ear)

For bitrate levels above this, I would leave lossy compression and go for lossless.
montython is offline   Reply With Quote
Old 23rd November 2007, 00:47   #9  |  Link
tebasuna51
Moderator
 
tebasuna51's Avatar
 
Join Date: Feb 2005
Location: Spain
Posts: 5,248
Quote:
Originally Posted by montython View Post
192 kbit/s is usually considered to be a bitrate level which ensures transparency. (not exactly the same with the original, but the differences are inaudible to human ear)

For bitrate levels above this, I would leave lossy compression and go for lossless.
Good post and well explained, only a little point:

Quote:
Originally Posted by fenomeno83
Bitrates 192-256-320-448 aren't necessary
High bitrates (192-320) are necessary for multichannel 5.1
tebasuna51 is offline   Reply With Quote
Old 23rd November 2007, 14:42   #10  |  Link
fenomeno83
Registered User
 
Join Date: Aug 2006
Posts: 330
THANKYOU!!

So,for stereo encoding:

If I must encode between 16 and 32 use aac he v2,44100 (but if original sampling is less,use the original)

If I must encode between 40 and 80 use aac he v1,44100 (but if original sampling is less,use the original)

If I must encode between 96 and 192(trasparence) is better to use aac lc (I will get a quality perfect!),44100 or 48000(better keep the original,but depends of uses and compatibility)

It's true?

And about application?
for me makes sense use aac lc(96 to 192)to encode cd or dvd concert
aac he v1(40 to 80) for film
and aac he v2?

and about 5.1 encoding?

Last edited by fenomeno83; 23rd November 2007 at 14:48.
fenomeno83 is offline   Reply With Quote
Reply

Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT +1. The time now is 18:47.


Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2017, vBulletin Solutions Inc.