missing the high freq in captured audio

[J37X]

Hi, my first post here. I'm having some trouble with loosing the high freq audio in my captured video. I'm capturing mostly music videos, so the audio just as important as the video quality. Anyway here's my setup, An ATI AIW 128pro using s-video input from my cable box, and an SBLive Value sound card with the audio going straight from the cable box to the sound card line-in. I'm using Virtualdub and capturing 640x480 using huffyuv, and PCM audio 44.100kHz, 16bit. Now when i listen to the audio in realtime it sounds fine, so i'm loosing it somewhere in the capture process, or the sblive card isnt able to record the full bandwidth of the audio input. I've tried the settings in the Creative mixer, and in the Sound properties page in the control panel, but maybe i'm missing something else, anyone have any ideas?
some other info on my system if needed:
Abit KG7, AMD tbird 1200@1470, 512MB pc2700, WinXP
thanks

[J37X]

...forgot to upload the graph showing the drop off

http://users.adelphia.net/~93zz/freq.jpg

[cjv]

1) Well the first thing I can suggest..for music videos anyways, but it doesn't help your overall problem, is to discard the captured audio altogether, and use a WAV or MP3 as the audio source for the video. (I realize that not all videos have the same music as the CD version..ie: extra talking, etc, so it won't always work)

2) You have a SBLive..try capturing at 48000 sample rate instead of 44100 (and use SSRC if you need to downsample later). The SBLive works best at that frequency. (This should make a difference..it should also keep sync better)

3) Use better quality cables, but that is probably not the problem as you say it sounds fine before capturing.

4) I believe changing the mixer settings in the volume properties will have no effect on anything if you are capturing direct from Line-In.

5) Does your cable box have spidf out? That way you will get a pure digital signal, but you must have the sblive with digital in (ie, not the OEM Value), or mod your current sblive.

6) Try a different capture app..you are probably going through the WDM wrapper with your ATI card. I find a WDM capture app (iuVCR, VirtualVCR) works and sounds great with my setup..and I have almost the same system and sound card as you.

6) If nothing else works, you may be able to find some good DirectX plugins for SoundForge/Cooledit to make it sound better.

cjv

[J37X]

cjv: I'll have to try the 48000, I didnt think it would make that much of a difference. Its funny you mention the spdif digital audio, i've been trying to get around to setting that up, I need to make a small circuit to convert the spdif signal to a TTL level signal so the sound card can accept it. And also your right, im using the WDM wrapper for the ATI, ill check out those other ones and give them try,,
thank alot

[^^-+I4004+-^^]

j37x>I'll have to try the 48000, I didnt think it would make that much of a difference.

sure,try that first:it should give more hifreq.
(although in theory 44khz should be enough,as it should be able to transfer freq. of up to 22khz.....
48 should then transfer up to 24khz,the only problem is where will you get audio signal of that quality.....)

also i wouldn't trust that nero wave editor spectrogram too much.....it's better to listen instead!

i noticed (on SB PCI128) that i have more hifreq.
on my win98 system than on win2k system
(i was ripping some LP records so i'm sure that win2k drivers are worse than win98 ones...)

[cjv]

Even though my SBLive captures the highs quite well, with a music video, the music is the most important. I've been capping a lot of vids recently from MTV, and I thought I'd share my method that works quite well and restores the CD-sound highs and punch. I know an audio engineer would just cringe :), about the order and especially with the compressor, but it works well for my purposes.

1) open 48000Hz, 16bit PCM wave in Soundforge
2) SonicFoundry NoiseReduction 2.0, capture noiseprint, reduce by 12db
3) Light compressor, any will work, (I used the built-in Wave Hammer), threshold -15db, ratio 2:1 (this restores the CD punch..without over-compressing)
3) normalize to 98% peak
4) second pass, noise reduction, 12db, high-shelf boost, 3db@7000 (eliminates any remaining hiss and then restores highs)
5) save as WAV
6) encode as Vorbis q6

Doing direct comparisons with a an --aps MP3 version of the song in question, and the resulting Vorbis (on my < $500 stereo), I cannot tell the difference. Hope some of this helps,

cjv