Newbie question

[Manngo]

Hello!

This is my first post in the audio forum. My problem, that I cannot solve myself:
Have an AVI with VBR MP3. want to encode to DVD with ac3.
Problem: volume is low, even in the source. plays normal when loudspeaker volume is set to 50% of max. (ideal would be 25%)
Additional info: I had to extract the VBR MP3 from the avi to wav, so I would encode wav to ac3, with increasing volume.
Already tried: measured average RMS level with SForge: 27,2. According to the guide in the sticky set the parameters of my ac3 encoder (Sonic Scenarist built-in). The result is the same volume as the wav and mp3. Even when played on standalone DVD player.

Also tried besweet GUI with normalize to 120%. Than wav become louder, but the encoded ac3 not.

So please help, how to do it? What do I wrong?

[Manngo]

hmmm...

usally no respone means that I ask a terribly stupid thing. And it is true. Volume should be in the middle of the scale. Between 0 dB and -60 dB. So having a ac3 file like this is perfect.
Do I see good? Someone could type a simple YES or NO. Would help in making me sure. Thanks in advance.

[KpeX]

Well if the WAV has correct volume but the AC3 is too low, it could be a playback problem. Also check if your DVD player/amp supports DRC or other AC3 features.

[Manngo]

Thank You for the ansear!

The ac3 volume is the same as the wav.

I tested the wav with Soundforge, and it is normailzed, the highest peak is 0 dB. Increasing the voulme couses loss at the louder parts.
I think the dinamic range is too wide, but it is not a matter.

What would you do in my place? Leave volume unchanged?

[Pyscrow]

Originally posted by Manngo
Thank You for the ansear!

The ac3 volume is the same as the wav.

I tested the wav with Soundforge, and it is normailzed, the highest peak is 0 dB. Increasing the voulme couses loss at the louder parts.
I think the dinamic range is too wide, but it is not a matter.

What would you do in my place? Leave volume unchanged?

You need to Compress it (or raise the compression).

[Manngo]

Thank you very much guys, I have managed with this problem.
I used Soundforge to normailze the wav to -16 dB, using average RMS. I just had to select 'Apply dynamic compression' when clipping accours, and check the 'Use equal loudness contour' checkbox. The audio is very comfortable to listen to now.

I could find a good level by experimenting and normaizing for 4 times with different target RMS level. Some analyzing program could, that can show information about loudness of a track on a statistical scale like this example:
__0dB-_-5dB_________1%
_-5dB-_-10dB________8%
-10dB-_-15dB_______11%
-15dB-_-AVG RMS____80%
_____________Sum: 100%

I mean the second column shows, how long part of the audio contains peaks within the range specified in the first column (knowing the average RMS). Using this information, determining the target average RMS level would be easy.

Do someone know a program with this features?

[LnxPeng]

Would a command-line program work? I am in the planning stages of writing a program that changes dynamic ranges in WAV and/or AIFF files. The analysis would be a necesary part, so I could start by making it spit out the information you want.

And if I get this program finished, how would you want it, executable for Win32 (or Linux x86) or source code (C)?

[Manngo]

What great news!

I think command line program is good, becouse it can be used in automatized processes easily, such as DVD2SVCD.
Concerning the format, I have Win2K, never had Linux. I'm just a simple user, so the executable would be the best for me.

I wish You a pleasant work and I'm looking for You program wery much!

[jorel]

newby question?
no!
good questions and answers in my point of view, great thread!
;)

@ LnxPeng
please, executable for Win32 will be amazing.
thanks in advance!
:)

[KpeX]

LnxPeng:

That sounds very interesting. Ultimately, when this program has passed past an informational stage, you could also consider making a plugin for BeSweet - example source is available on DSPguru's site (http://dspguru.notrace.dk/plug-ins.htm). This would make the dynamic range changing ability available for a wide range of conversions.

[LnxPeng]

Program is rolling along, but I did run into a math problem (not critical) dealing with conventions used with PCM data:

PCM, as I understand, is represented as a signed 16-bit integer. This means that the valid range is from -32,768 to 32,767 (and I have confirmation that WAVE files use this range). This means that one side or the other won't be able to achive exact 0 dB, since I will have to reference either 32767 or 32768 as 0 dB.

It's no big deal, but I would like to have it output the exact same results as the other programs listed.

For now, I will treat 32767 as reference at 0 dB, and any -32768 an if statement will catch and set equal to 0 dB also.

Or I could just add one to the sample to calculate RMS on the negative side.

Anyways I've got an algoritm that gets close. I've got to add in the WAVE file handling. Should I post a link or do I need to attach the archive here somehow?

[LnxPeng]

KpeX:

Thanks for the link. Right now I'm using a port of GCC, so I'm not sure if I can make a plugin with it. I'll definately look into it once I get the algorithm smoothed out.