Additions and comments on "GUIDE LIST: Stereo-to-surround Conversion.."

[raquete]

off topic
Elektra,
can you host only one track from the CD 16/44100 PCM WAV, (Evanescence - The Open Door)?
i (still) don't use the tools you're using and i want to do 5.1 and host for you again,then...you download,compare and tell me if sounds good!
can you?
regards!

[Elektra999]

Hi, Steve.
I believe that it is possible to make 6.1 with WaveLab v5.

To install AsioDrivers.
Menu / Tools / Preferences / Audio Card / Playback/Record, to select AsioDrivers.
If we select / Connections, we will see the channels and the order, OK.

http://img74.imageshack.us/img74/7447/asiodriversee7.jpg

01. Menú / Edit / Audio Properties, 32bits float, OK.
02. Menú / Eliminate DC offset, OK.
03. Menú / Edit / Create Audio Montage from Wave
04. Whole Range, OK.
05. Clone 4 tracks “stereo”. Menu / Edit / Mode / 8 Channels.

http://img324.imageshack.us/img324/7459/61canalesrj0.jpg

http://img60.imageshack.us/img60/23/014pistasmh4.jpg

http://img60.imageshack.us/img60/5149/024pistasav5.jpg

06. Track 1, Effect FL FR. Track 2, Effect CLFE. Track 3, Effect SL SR (channel ES Matrix). Track 4, Effect SL SR .
07. Channel 1 – FL. Channel 2 FR.

http://img261.imageshack.us/img261/4319/01fe4.jpg

08. Channel 3 – Center. Channel 4 – LFE

http://img317.imageshack.us/img317/5369/02tz2.jpg

09. Channel 5 – Surround (ES Matrix). Channel 5 – Surround (ES Matrix).

http://img410.imageshack.us/img410/1848/03rt7.jpg

10. Channel 6 – SL. Channel 7 – SR.

http://img108.imageshack.us/img108/3554/04op3.jpg

Steve. It seems to me that, is mixed well, channel (Central back) with WaveLab?

Thank you for his Suite V.I:)

Frank

[Elektra999]

11. Render, save Multimono.

http://img442.imageshack.us/img442/6532/guardarij7.jpg

WAV 1 - CHANNEL FL
WAV 2 - CHANNEL FR
WAV 3 - CHANNEL CENTER
WAV 4 - CHANNEL LFE
WAV 5 - CHANNEL ES MATRIX?
WAV 6 - CHANNEL SL
WAV 7 - CHANNEL SR
WAV 8 - DELETE

[Elektra999]

Hi, Steve and to all.
I believe that the 6,1 obtained with WaveLab 5, have come out well.
I have a question, so that 32bits float is applied?

Thanks
Frank

[ursamtl]

Elektra,

Sorry but I do not understand your question. If it's about bit depth, the best advice I can give is to always work in the highest bit depth you have available, preferably 32 bits, and then reduce to 24 bits for DVD or 16 bits for CD. Whenever reducing, be sure to use dither. In the past I had read that dithering when reducing from 32 to 24 bits is unnecessary, but recently I read some that to maintain the maximum resolution, always dither when reducing bit depth. However, avoid doing this until the finishing your project.

[Elektra999]

I,m sorry Steve
My question was:

01.In order that to apply them 32bits?
02. In an audio of 24/96000 is it necessary to apply them 32bits to codify 5.1?

I am glad to see him Steve :)
Thanks

[ursamtl]

1. Convert everything to 32 bits for processing.
2. Create the 5.1 mix.
3. Convert the individual files to 24 bits with dither. If this is not practical, you can do without dither when moving from 32 to 24 bits but you should never convert to 16 bits without dither as the results (quantization distortion) could be audible in quiet passages or in the surrounds.

[Elektra999]

Thank you very much Steve.

I believe that with its guide WaveLab 5.1, it is possible to add 3 effect, the dithering UV22 HR to every track stereo. From 32/96000 to 16/44100.

Regards! :)

[Elektra999]

Hi Steve
Do you know some VST plug-in Surround Panner of 6/7 channels that it works in Plogue Bidule? Though of payment.

I am waiting for the update of the DTS Master Audio, I will have to think of doing since 7.1

6.1 ES MATRIX I have managed it to do, as I commented in this post mas above .

Thanks Steve
Frank

[ursamtl]

Hi Steve
Do you know some VST plug-in Surround Panner of 6/7 channels that it works in Plogue Bidule? Though of payment.

I am waiting for the update of the DTS Master Audio, I will have to think of doing since 7.1

6.1 ES MATRIX I have managed it to do, as I commented in this post mas above .

Thanks Steve
Frank

Hi Frank,

You could try looking at http://acousmodules.free.fr/. This is a French site but there is also an English section. There are quite a few interesting multichannel panning plugins available.

[Elektra999]

Thanks Steve
I have been experimenting with the plug-in SpatMass 66. It is is more or less what was searching.

http://img216.imageshack.us/img216/2210/dibujo2nc0.jpg

I believe that the connections go this way:

http://img299.imageshack.us/img299/1948/dibujo3oz2.jpg
CS - SR - FR - LFE - CENTER - FL - SL

Audio Fle Recorder

WAV 0 - FL
WAV 1 - FR
WAV 2 - CENTER
WAV 3 - LFE
WAV 4 - SL
WAV 5 - SR
WAV 6 - CS
WAV 7 - DELETE

Regards :)

[Elektra999]

Hi Steve.
I have been following step by step, its guide of Nuendo. I have to say, that it seems to me fantastic the guide and her work. But I have two questions:

01. The Surround Panner I have fit it thus and not if it is well?

http://img92.imageshack.us/img92/4002/dibujoms3.jpg

02. In Output 5,1, I do not have the OCtoMaxx2-0, but the Pick Masters. The connections I have made them, but they seem to me thus
http://img283.imageshack.us/img283/9725/dibujo2zi2.jpg

I have used the Steniberg Cubase SX 3, and the mixture in 6 channels I believe that to come out well.

Thaks Steve
Frank

[ursamtl]

Your peakmaster is only affecting the front channels. If you only have two channels to limit, the most important ones in this process are the rears because the surround generation algorithm tends to unmask peaks and transients in the material that are normally cancelled out in the regular 2-channel stereo.

[Pookie]

I'm probably not looking in the right area - wondering if there is a recommended free VST host for this guide. Isn't Bidule no longer free? Perhaps Kristal Audio Engine instead?

[Elektra999]

Where I can find, the plug-in Octomaxx or PeakMaster of 6 channels?

Thanks

[ursamtl]

I'm probably not looking in the right area - wondering if there is a recommended free VST host for this guide. Isn't Bidule no longer free? Perhaps Kristal Audio Engine instead?

Kristal is a nice program but only does two channels at a time and only in real time. There is a guide for using it with the V.I 2-channel plugins that a guy on videohelp.com wrote (I listed it in the surround guide sticky at the top of this forum).

Plogue released a new demo on November 13. This will work until mid-February 2007. You can also try this: GUIDE: Converting stereo to 5.1 surround for FREE (http://forum.doom9.org/showthread.php?t=105684)

Regards,
Steve.

[ursamtl]

I believe Octomax came with Nuendo 2. I don't know if there is a 6-channel version of Peakmaster. You could set up three 2-channel limiters in the bottom three effects slots in Nuendo. Just move the connections until you line them up the way you want.

[Elektra999]

Thanks Steve and Pookie

I believe that you say to me, that she connects therefore the PickMaster, is thus?

http://img272.imageshack.us/img272/7388/dibujojt4.jpg

It completes question: the order the channels with the PeakMaster 2 channels, is thus?
WAV 1 - FL
WAV 2 - FR
WAV 3 - CENTER
WAV 4 - LFE
WAV 5 - SL or SR.?
WAV 6 - SR or SL?

Thanks you very much

[Pookie]

Thanks for the info, Steve :)

[jordisound]

Hi, i've a question about VI base method guides. I want to use VI plug-in with my audio editor. May I select audio 32-bits or 16-bits?

[ursamtl]

Hi jordi,

Yes, V.I will work with whatever bit depth you set for your audio editor. However, it's a general rule for audio editing that you should always work in 32 bits and then convert down to 16 bits as the last step and add dithering when you do the conversion. By the way, if you're doing the sound for DVD, you can reduce to 24-bit depth instead of 16. The quality is better than 16 and the DVD spec fully supports 24-bit audio up to a 96kHz sampling rate.

Regards,
Steve.

[Elektra999]

Hello
Steve, ¿if audio the original one, is to 16bits/48000. .hz don't mention it serves, to talk to 24bits/96000…hz?

If audio the original one is, to 24bits/96000 ¡sounds of wonder! and z sounds better than to 16/48000…

[ursamtl]

Franck,

If I understand correctly, you're basically asking if there's any benefit to upsampling to 24/96 if the source is 16/48. I'd say yes in my experience it helps. First of all, audio apps with VST plugins end up converting the data to 32-bit resolution anyway so the bit depth is automatically increase (and must be dithered if reduced, even to 24-bit according to pros such as Bob Katz). As for upsampling to 96kHz, I've been reading some material lately about the effect of filtering on data. Audio data at 44.1 and even 48kHz can suffer from filter rippling effects in the highs, whereas 96kHz audio moves the filtering up far beyond the limits of human hearing, so it can help. I know from experience with my home CD/DVD player (A Yamaha DVDS657), its CD upsampling function tends to make high frequency transients such as cymbals, guitar strums, etc., sound much smoother to my ears.

The only time I'd stick at 48kHz is if I were working with a slower PC or had very limited storage resources. But again, the golden rule is to never, never, never work at a 16-bit resolution. If you need the final file to be at that resolution reduce as the very last step and with dither. Even a level change after dithering can result in data loss.

[jordisound]

What happens if the source is mono? I've extracted audio from a AC3mono DVD. I convert mono wav to stereo wav (both channels are identical) to apply VI plug-in. Does VI plug-in works OK?

[ursamtl]

No, a mono source won't work because V.I doesn't synthesize fake ambience. It simply extracts the ambience already present in stereo signals using modified ambisonic/M-S principles. Just duplicating the mono file on two channels won't give you the difference information required to create the side channel in M-S or the Y channel in ambisonics.

I've had some success with mono material by running it through a convolution plugin first with some carefully chosen reverb impulses loaded. You might also try the freeware reverb plugins Ambience or Glaceverb. There are also some good stereo simulation plugins available. Elevayta's Wide Boy is a good one (it's almost free, just a 90-cent US bandwidth fee). I'd stay away from some stereo simulator plugins such as Voxengo's Stereo Touch as it's designed for smooth keyboard-pad style sounds (according to their web site). Anything with sharp transients such as percussion sounds really terrible through this type of simulator. The free mda plugins have a stereo simulator plugin that's not bad. you might try it.

Anyway, good luck.
Steve.

[ursamtl]

Hi folks,

I frequently get private messages with questions about using V.I. I don't mind answering them (and I think my track record on answering is ok), but I think it would be better if everyone asked their questions in the public forum in this thread. It would help others if you'd ask the questions in the forum instead of privately. Like most forums, Doom9 is a community where we all improve our hobby by sharing knowledge. In addition, I (or whoever else you message privately) may not have the answer you're looking for or perhaps I'm out of town, etc.). Someone else may be able to help you.

One question I get a lot is when someone takes a file that was previously compressed (usually 2-channel AC3, MP2, MP3) and tries to create a surround file using V.I or one of the other methods. After converting, artifacts or weird sounds seem to appear that didn't seem to be there in the stereo file.

I my experience, it's always tricky to try creating surround out of files that were compressed (AC3, MP2, MP3, etc.). V.I works by extracting ambience that was already in the track. So does Dolby Pro Logic-based methods, Ambisonics, Quadraphonic methods, etc. The compression algorithms used by these formats (AC3, MP2, MP3) all mess the ambience in some way or another. You don't really hear this much when the files are in stereo the algorithms were designed so that the stereo files mask these artifacts. As soon as you try to extract the ambience, you unmask the artifacts.

So whenever possible, start with an uncompressed source, such as the 2-channel LPCM from a movie, or a wave file.

If you do have a compressed file, I suggest redigitizing it. This means, recording the analog output of the stereo file back into your PC. This way, the playback equipment recreates a version of the soundwave and may reduce the artifacts problem.

So remember, share your ideas. I don't mind receiving private messages, but I often think it's too bad we didn't share the problems and solutions in public where more people could benefit.

Regards,
Steve.

[jordisound]

it's a general rule for audio editing that you should always work in 32 bits and then convert down to 16 bits as the last step and add dithering when you do the conversion. By the way, if you're doing the sound for DVD, you can reduce to 24-bit depth instead of 16. The quality is better than 16 and the DVD spec fully supports 24-bit audio up to a 96kHz sampling rate.
Hi Steve,
I've a question about what you said.
Using Aften to encode from the fRfL, CLFE anf sLsR wav (32bits 48000hz), the ac3 final is 16bits 4800hz. If Aften converts 32 to 16 while encoding, maybe is not necessary convert the wavs to 16-bits as you recommended at last step. isn't necessary?

[tebasuna51]

Hi Steve,
I've a question about what you said.
Using Aften to encode from the fRfL, CLFE anf sLsR wav (32bits 48000hz), the ac3 final is 16bits 4800hz. If Aften converts 32 to 16 while encoding, maybe is not necessary convert the wavs to 16-bits as you recommended at last step. isn't necessary?

AFAIK Steve speech about LPCM for 24 bit 96 KHz.

Ac3, internally, don't have different sample size like wav and the input wav can be with any sample size. The first thing Aften do is convert any sample size to 32 float, is not necessary any previous reconversion.

"...the ac3 final is 16bits 4800hz."?

For what do you think this? Maybe is your decoder, try with NicAc3Source() in AviSynth, the output is always 32 bit float. Use SoundOut or last BaHappyMod to see.

[raquete]

@ tebasuna
BaHappyMod? i don't know.where to find? (in behappy thread? )

For what do you think this?hummm..the same here tebasuna.
loading one wave(6 interleaved) 48k-32bit in EncWavtoAC3 give me the ac3 final is 48k-16bit.

[ursamtl]

I strongly recommend reducing bit depth to 16-bit resolution with dither (dither is ESSENTIAL when going to 16-bit) before encoding. Use triangular dither. Noise shaping might be a good idea but lossy compression may filter out some or all of the dither noise added when using noise shaping.

If Aften does the bit-depth reduction with dithering then yes, it could be done by it. If it's doing it by truncation (chopping off data), then do not let Aften do the reduction, do it yourself beforehand.

By the way, slightly related to this is sample rate conversion. With the availability of equipment that supports higher sampling rates such as 96kHz and 192kHz, converting down to 48kHz (DVD) or 44.1kHz (CD, whether regular audio or surround), requires good conversion. Here's an interesting comparison of sample rate converters I've discovered recently:

http://src.infinitewave.ca/

Looks like Cool Edit Pro/Adobe Audition provides some of the better performance among tools used by a lot of forum users. r8brain isn't as impressive as I expected, and some of the others (Wavelab for example) are a bit disappointing.

[raquete]

(i'm lucky,i use audition :) )

thanks for the clarifications and the picture Steve!

[tebasuna51]

If Aften does the bit-depth reduction with dithering then yes, it could be done by it. If it's doing it by truncation (chopping off data), then do not let Aften do the reduction, do it yourself beforehand.
Read my post:
"The first thing Aften do is convert any sample size to 32 float, is not necessary any previous reconversion."

No bit-depth reduction, no truncation.

@Raquete, Jordisound
What software say you an ac3 is 16 bit depth?
Where are the fields in ac3 header about bit depth?

We only can say bit depth accepted by a encoder or supplied by a decoder, internally not exist difference.

@raquete
Yes BeHappyMod is in behappy thread.

[raquete]

The first thing Aften do is convert any sample size to 32 float, is not necessary any previous reconversionall right,is clever!
What software say you an ac3 is 16 bit depth?in ac3filter playing AC3 (inside mpclassic) but can be my wrong configurations. to tell the true i don't listen the files in pc,i only play in mpclassic to see the infos.
what tool can be used to read the complete AC3 infos?

:thanks:

[tebasuna51]

in ac3filter playing AC3 (inside mpclassic) but can be my wrong configurations.
In Ac3Filter config Main tab -> Output, you can select the bit depth than this decoder can supply 16/24/32 bit int or 32 float.
what tool can be used to read the complete AC3 infos?

I make a simple header ac3/mp3 reader (http://www.mytempdir.com/1093720) (only first frame showed and statistics).

[raquete]

great! :)
adjusting ac3 filter give me this results in decoder info:
Input format: PCM32 3/2.1 (5.1) 48000
User format: PCM32 3/2.1 (5.1) 0
Output format: PCM32 3/2.1 (5.1) 48000

Decoding chain:
(PCM32 3/2.1 (5.1) 48000) -> Processor -> (PCM32 3/2.1 (5.1) 48000) -> Dejitter -> (PCM32 3/2.1 (5.1) 48000)

Filters info (in order of processing):

Processor:
(PCM32 3/2.1 (5.1) 48000) -> PCM->Linear converter -> (Linear PCM 3/2.1 (5.1) 48000) -> Input levels -> (Linear PCM 3/2.1 (5.1) 48000) -> Mixer -> (Linear PCM 3/2.1 (5.1) 48000) -> Bass redirection -> (Linear PCM 3/2.1 (5.1) 48000) -> AGC -> (Linear PCM 3/2.1 (5.1) 48000) -> Delay -> (Linear PCM 3/2.1 (5.1) 48000) -> Output levels -> (Linear PCM 3/2.1 (5.1) 48000) -> Linear->PCM converter -> (PCM32 3/2.1 (5.1) 48000)

Dejitter:
-

I make a simple header ac3/mp3mytempdir always give me "The file is temporary unavailable, please try again later."
is possible to host in rapidshare?

thanks so much tebasuna!

[jordisound]

@tebasuna51: did your application (LeeAudBi) show information about bitdepth on AC3?

[tebasuna51]

Try here.


Edit: New link and version LeeAudBi03b.7z

Edit: Link dead, attached only exe file

Edit: new version here http://forum.doom9.org/showthread.php?p=1522330#post1522330

[jordisound]

Well, I've tested again with bepipe and aften in commandline, using a script to merge flfr, clfe and slsr (tebasuna did it and send me)

global OPT_AllowFloatAudio=True
f = WavSource("C:\rip\FLFR.wav")
c = WavSource("C:\rip\CLFE.wav")
s = WavSource("C:\rip\SLSR.wav")
MergeChannels(f, c, s)

and tebasuna was right.
http://img11.imagepile.net/img11/81992aften32.png


@tebasuna51: did your application (LeeAudBi) show information about bitdepth on AC3? I can't find this field.

[tebasuna51]

@tebasuna51: did your application (LeeAudBi) show information about bitdepth on AC3?
One more time?
Not exist different bitdepth in ac3 the data info is in frequency domain, not in time domain, then the bitdepth is not applicable.

I say all bitdepth's samples are translated to 32 float before the time -> frequency conversion and the precision is defined by the bitrate.

[ursamtl]

Yes, but do you not agree that the data has to be converted back into the time domain to be sent to the D/A converter? WinDVD for example, will look at an AC3 stream in a DVD file and identify it as 16-bit/48kHz. Is it taking 32-bit data and truncating it to 16?

[raquete]

curious...encoder adjusted to 448 and 512:

(File to analyze: H:\ROBERT~1\AC3512~1\01LOVE~2.AC3
SampleRate 0 : 48000 KHz.
BitRate 15 : 448 Kb/s.
...and...
File to analyze: C:\DOCUME~1\ADMINI~1\DESKTOP\01LOVE~1.AC3
SampleRate 0 : 48000 KHz.
BitRate 16 : 512 Kb/s.

why i'm getting this results?

[jordisound]

Sorry, tebasuna51, i didn't undestand your first answer. Now, I do.
I was using wrong software to reasd the ac3 info, that software show 16bits, when it doesn't exist in AC3!

[tebasuna51]

Yes, but do you not agree that the data has to be converted back into the time domain to be sent to the D/A converter?
Of course, but this is a decoder problem.
WinDVD for example, will look at an AC3 stream in a DVD file and identify it as 16-bit/48kHz.
Impossible. Nothing in the ac3 stream header say the original sample size was 8/16/24/32 or float.
A good decoder can be instructed to decode from frequency to time using like output 16/24/32 ...
Is it taking 32-bit data and truncating it to 16?
The data in the ac3 stream are exponents and mantissa frequency values and is most precise with more bitrate.

Really I don't understand the problem. We have 3 stereo wav's (flfr, clfe, slsr) 32 bit float, send the data directly to Aften with:
global OPT_AllowFloatAudio=True
f = WavSource("C:\rip\FLFR.wav")
c = WavSource("C:\rip\CLFE.wav")
s = WavSource("C:\rip\SLSR.wav")
MergeChannels(f, c, s)
and verify if is incompatible with DVD or any soft identify this ac3 like 32 bits.

[ursamtl]

According to the book I'm reading right now, Ken Pohlmann's The Principles of Digital Audio (an excellent "bible" recommended by engineer Bob Katz in one of his articles on www.digido.com), AC3 in principle supports up to 24-bit audio, not 32 bits. He also states that the AC3 decoder outputs linear PCM after the decoding. So you're right, my mention of the data being converted back into the time domain before D/A conversion is a decoder issue. The question here is whether inputting 32-bit audio to Aften will output 32-bit audio to the decoder, which will in turn output 24-bit audio. Pohlmann also mentions a dither flag in the bitstream. What do you know about this? Does Aften support the dither flag?

[raquete]

AC3 in principle supports up to 24-bit audio, not 32 bits.i read in some Dolby.inc and ATSC docs that AC3 support 16,18,20 and 24 bits,have some data sheets showing that too.
months ago i asked about AC3 24 bit in the Aften thread and why not we don't was using this feature ....no clever answers.
now i have 3 results from tebasuna's LeeAudBi log file using one interleaved 6 channels 32 bit from EncWavtoAC3 encoder:
1- BitRate 15 : 448 Kb/s.
2- BitRate 16 : 512 Kb/s.
3- BitRate 18 : 640 Kb/s.

no one result is 32 bit and it send me to another question:
why 15,16 and 18bit? :confused:

thanks for any clarification.

[foxyshadis]

Internally it's all infinite bitdepth, because it's all in the frequency domain. It'll be constrained by the filterbank construction and the bitrate, but in theory if you could throw enough bitrate at it you could construct an output that more closely matches a 32-bit input than its 24-bit downsampling. Realistically there probably isn't enough resolution in the bottom filterbanks, that's a problem in mp3 too; maybe eac3 handles it better, like aac or vorbis.

Those three specific bit depths are probably what the codec considers the maximum that it can reliably produce. But then again, you'd probably have to ask the EncWavtoAC3 author.

It would be interesting to renormalize a 24-bit track to 32-bit -140dB and find out whether the codec really is capable of 32-bit, in the absense of other factors.

[tebasuna51]

now i have 3 results from tebasuna's LeeAudBi log file using one interleaved 6 channels 32 bit from EncWavtoAC3 encoder:
1- BitRate 15 : 448 Kb/s.
2- BitRate 16 : 512 Kb/s.
3- BitRate 18 : 640 Kb/s.

no one result is 32 bit and it send me to another question:
why 15,16 and 18bit?
These values aren't bits per sample, but BitRate codes':
BitRatecod Bit Rate
------------- --------
'00000' ( 0) 32 kbps
'00001' ( 1) 40 kbps
'00010' ( 2) 48 kbps
'00011' ( 3) 56 kbps
'00100' ( 4) 64 kbps
'00101' ( 5) 80 kbps
'00110' ( 6) 96 kbps
'00111' ( 7) 112 kbps
'01000' ( 8) 128 kbps
'01001' ( 9) 160 kbps
'01010' (10) 192 kbps
'01011' (11) 224 kbps
'01100' (12) 256 kbps
'01101' (13) 320 kbps
'01110' (14) 384 kbps
'01111' (15) 448 kbps
'10000' (16) 512 kbps
'10001' (17) 576 kbps
'10010' (18) 640 kbps
the binary data in ac3 header.

From Dolby doc "ATSC Standard: Digital Audio Compression (AC-3)":

"8.2.1.1 Input word length
The AC-3 encoder accepts audio in the form of PCM words. The internal dynamic range of AC3 allows input word lengths of up to 24 bits to be useful."

Maybe this is the origin of the 24 bit depth of ac3 streams, but like foxyshadis say is not inherent to the format (infinite bitdepth), is a feature of the encoder. Aften accept 8/16/20/24/32 bit int and 32/64 bit float wav's.

@ursamtl
Yes, Aften construct the DitherFlag in ac3 streams. But I don't know if this can clarify:

"7.3.4 Dither for Zero Bit Mantissas (bap=0)
The AC-3 decoder uses random noise (dither) values instead of quantized values when the number of bits allocated to a mantissa is zero (bap = 0). The use of the random value is conditional on the value of dithflag. When the value of dithflag is 1, the random noise value is used.
When the value of dithflag is 0, a true zero value is used. There is a dithflag variable for each channel. Dither is applied after the individual channels are extracted from the coupling channel.
In this way, the dither applied to each channel's upper frequencies is uncorrelated."

[raquete]

These values aren't bits per sample, but BitRate codes'all right,is clever.
if the interleaved wave source is 16,24,or 32 bit,after encode with aften the result is 32 bit.
the ac3filter is showing the right bit of the input!
thanks so much.
:-)

[tebasuna51]

all right,is clever.
if the interleaved wave source is 16,24,or 32 bit,after encode with aften the result is 32 bit.
Nope. If is a 48 KHz. the result is an ac3 stream with:
Stereo 6 channels
Bit Rate Average bits/sample Average bits/sample
-------- ------------------- -------------------
32 kbps 0.333 0.111
40 kbps 0.416 0.138
48 kbps 0.5 0.166
56 kbps 0.583 0.194
64 kbps 0.666 0.222
80 kbps 0.833 0.277
96 kbps 1 0.333
112 kbps 1.166 0.388
128 kbps 1.333 0.444
160 kbps 1.666 0.555
192 kbps 2 0.666
224 kbps 2.333 0.777
256 kbps 2.666 0.888
320 kbps 3.333 1.111
384 kbps 4 1.333
448 kbps 4.666 1.555
512 kbps 5.333 1.777
576 kbps 6 2
640 kbps 6.666 2.222

Is called compression :rolleyes:

After a decoder can decode the ac3 stream to 8/16/24/32/64 or the bitdepth desired.

[raquete]

Nope. If is a 48 KHz
lol.i'm sorry,i really don't understood what you mean in the last post....
you mean that if the interleaved wave is not 48k can be different? :confused:
i told you that when i load one interleaved(wav 6 channel)48k - 16 to 32 bit in Aften i have always the ac3 in 32 bit.
where i'm lost?