44,100 to 48,000 sampling

[entwood]

*This is also a VCD to DVD question but I'm only interested in the audio.

I've got 24 mpeg1 files with layer2 audio that I want to put on several DVD's.

Is there a correct way to convert the audio to 48,000 hz. I want to keep the audio quality at least the same.

If I'm not careful though I'll probably rencode giving me worse audio at a higher bitrate.

thanks.

[gastorgrab]

SSRC (Shibata Sample Rate Converter) is freeware, and accepts *.WAV input.

You can find it in the Tools Section, but its command line.

There is another version at Rarewares.org that includes a simple GUI with SSRC.exe.

I would suggest that you use "SSRC_HP.exe" (just rename to SSRC.exe), instead of "SSRC.exe" if you want better quality. Both versions are included in both downloads.

Set as follows for pretty descent results;

[wulf109]

Headac3he may be a better choice. Comes with a GUI. Converts between formats in addition to re-sampling.

[gastorgrab]

Headac3he may be a better choice. Comes with a GUI. Converts between formats in addition to re-sampling.

I just tried to check it out myself, and the page is down.

Unfortunately my space provider Tripod had a HD crash, so this site will not be available for (hopefully only) a few days.

Is there another source?

[johns0]

Try ffmpeg with gui,it will convert 44khz mp2 to 48khz mp2 or ac3.

[entwood]

you know ffmpeg sound like an idea.

I'd like to find a program that can process movies, leave the video alone and convert the audio. I use Tmpg Enc for mpeg but it looks like it will always rencode the video. When I demux -> convert audio -> remux, the audio is out of sync.


btw John, that last line of yours might be racist.

[johns0]

btw John, that last line of yours might be racist.

Why would it be racist?Its a joke on his intelligence,nothing else.

[garrymum]

Just use TMPGenc, just tick audio only on the right hand side.

[zedolfblack]

I use WinAMP.

First i use VirtualDub to extract the audio from a video - i extract it in it's original format - usually MP3.

Then i load the audio track in WinAMP.
WinAMP preferences enables you to select the Disk Writer plug-in for Output.
Set the Disk Writer plug-in to save a 48000Hz stereo WAV and hit play.

I used to use Sound Forge but it took AGES.

Martin.

[lordsmurf]

SoundForge will do the best job quality-wise, but you can use SSRC in Besweet (use the GUI).

[junkmalle]

VirtualDub can convert audio sampling frequency. And it's free.

[entwood]

John01,
About my comment, keep it if you want. But I think part of the humour comes from Ben being black and denying that he's a theif. But again, if no one says anything, obviously no one if offended. Sorry My Bad.

Everyone
I've tried alot of your suggestions but the audio keeps going out of sync, when I convert to 48,000. I've even rencoded the entire video in Tmpg Enc.

[dphirschler]

Don't forget AVIsynth. Also free. The beauty with this method is you can work it right into your script instead of using an outside program. You can also test for sync immediately in Virtual Dub.

The command:

SSRC(48000)

Darryl

[entwood]

Ok, thanks for the help but right now all I'm doing is creating out of sync audio with these programs.

Since it keeps happening, I think there must be another problem with mpeg1 video that I don't know about.

BTW: Are any of you working on commission of some kind?

[DrWafer]

use any prog you like.

but first just get rid of any elecard mpeg codecs you have in your system.

99% of audio sync problems are due to this codecs.

[entwood]

I don't have a clue about these Mpeg codecs. My guess is that I have several on my computer and my programs just pick one.

How do I fix that up. Is elecard a typo

[lordsmurf]

BTW: Are any of you working on commission of some kind?

Yes. I get $0/hour, no benefits. Overtime is twice that.

[entwood]

I think I'll look for threads about mpeg codec and "elecard". Thanks for all the help. It's appreciated.


BTW- Just so you know I don't just ask questions. Problem is the info I put on the Internet has nothing to do with digital video and is only useful if you live by Niagara Falls.

[dphirschler]

Yes. I get $0/hour, no benefits. Overtime is twice that.

You get double overtime?! I want a raise!


Darryl

[mpack]

> SSRC (Shibata Sample Rate Converter) is freeware, and
> accepts *.WAV input.

It may be free, but I was very disappointed with the quality of its output. I had a VCD->DVD conversion project I was doing, and the audio track included a phone ringing... I don't know what kind of FIR filtering SSRC was doing to the audio, but the clipping and other distortion when that phone rang (which was unfortunately quite often) was very offputting.

It so happens I'm a programmer, so I eventually got around the problem by writing my own sample rate converter based on the simplest possible technique: insert or delete samples at regular intervals until the rate is correct. Before starting I kind of assumed this would be a failed experiment (otherwise why would people need something like SSRC), but in fact it worked perfectly, without SSRC's distortion. I imagine you wouldn't want to make a HUGE change to the sample rate that way, but for 44.1->48kHz it seems to work fine. Sorry I don't have to web site to host my tool on, but I've since discovered that there are others that work the same way.

[gastorgrab]

It may be free, but I was very disappointed with the quality of its output. I had a VCD->DVD conversion project I was doing, and the audio track included a phone ringing... I don't know what kind of FIR filtering SSRC was doing to the audio, but the clipping and other distortion when that phone rang (which was unfortunately quite often) was very offputting.

Have you tried SSRC in "two-pass" mode?

Try this:
ssrc.exe --rate 48000 --twopass --dither 1 --bits 16 --pdf 1 before.wav after.wav

SSRC default settings are intended to be fast, not efficient.

[mpack]

Well, thank for taking the trouble to offer the tip, but I already have a solution that is fast and gives perfect results, so why bother with anything fancier?

Besides, it's handy having my own .wav tool, because I can tweak it as needed to do other things in the same pass, eg. a slight tweak and it can adjust the length of the audio as it goes, to match the video.

I just have this sneaky suspicion that after all the fancy floating point filter stuff is done in SSRC, in the end it must quantize the result to the nearest integer... and maybe end up doing no better (and often worse) than I do with my simple trick.

[gastorgrab]

Rounding to the nearest integer, really?

I always assumed it multiplied each sample by some whole number to get to the lowest common denominator, so it could then divide by another whole number. It just seemed the logical way to do it.

I really don't know what the strategy of any of the Sample Rate Conversion apps are!

I know what your saying about SSRC. It will clip every time without two-pass set. I use it mostly with TMPGEnc, but i altered the file "TMPGEnc.acf" to include twopass settings.

[info]
toolame*.exe=tooLAME
LAME.exe=LAME
SCMPX.exe=SCMPX

WaveFS44.exe=WaveFS44
SSRC.exe=SSRC
SSRC_HP.exe=SSRC_HP


[SSRC:info]
Name="SSRC Standard"
SupportMethod=resample
Resample.InterfaceType=CommandLine
Resample.SupportInput=all
Resample.SupportOutput=all
Resample.SupportChannels=all

[SSRC:resamplearam]
" --rate %s":SamplingFrequency
" --bits 16"
" --dither 1"
" --pdf 1"
" %s":SrcFileName
" %s"estFileName


[SSRC_HP:info]
Name="SSRC Hi-Precision"
SupportMethod=resample
Resample.InterfaceType=CommandLine
Resample.SupportInput=all
Resample.SupportOutput=all
Resample.SupportChannels=all

[SSRC_HP:resamplearam]
" --rate %s":SamplingFrequency
" --bits 16"
" --twopass"
" --dither 1"
" --pdf 1"
" %s":SrcFileName
" %s"estFileName

[entwood]

Hey, your altering the settings for Tmpg. Where do you guys learn this stuff.

[mpack]

I always assumed it multiplied each sample by some whole number to get to the lowest common denominator, so it could then divide by another whole number. It just seemed the logical way to do it.

Well, there you are describing fixed point integer arithmetic. That's more efficient on some hardware, but in terms of accurancy it produces the same result as using floating point followed by a truncate to integer op.

It's been a while since I barely glanced at the SSRC source code, but ISTR the core was an ordinary floating point FIR filter.

I really don't know what the strategy of any of the Sample Rate Conversion apps are!

Well, you know how mine works now!