I've been using MAD with an M Audio Audiophile 2496 card for a week now. Here's what I've found.
The oscilliscope and VU meter is useless in 24 or 32 bit resolution mode; it just constantly squiggles, even during silent passages. With my particular card, 24 bit resolution only works with a waveOut buffer of 9000 ms or more. If I use a lower buffer, the MP3 files will start "screeching", making nothing but a loud static noise if I scan through the song. However, 32 bit resolution works fine. MAD's 32-bit output is padded 24-bit, correct Rob? Well, I'm using it with my 24-bit sound card, and it works perfectly with the default waveOut settings.
And like I even need to say this, MAD at 24-bit resolution blows Fraunhofer or audioEnlightenment out of the water. I'm using a Creek OBH-11 headphone amplifier mated to Sony MDR-V6 Studio Monitor headphones, and the difference is pretty stark and obvious. The MAD output sounds more natural, has better instrument decay, has more bass power, and the bass is smoother and the treble clearer. Fraunhofer sounds positively fake in comparison.
Regards to you all,
David
Hi,
On Wed, 29 Nov 2000, David Shin wrote:
I've been using MAD with an M Audio Audiophile 2496 card for a week now. Here's what I've found.
The oscilliscope and VU meter is useless in 24 or 32 bit resolution mode; it just constantly squiggles, even during silent passages. With my particular card, 24 bit resolution only works with a waveOut buffer of 9000 ms or more. If I use a lower buffer, the MP3 files will start "screeching", making nothing but a loud static noise if I scan through the song. However, 32 bit resolution works fine. MAD's 32-bit output is padded 24-bit, correct Rob? Well, I'm using it with my 24-bit sound card, and it works perfectly with the default waveOut settings.
For the oscilloscope (and VU meter), it's normal. MAD plug-in send 24 (32) bits and the visualization system doesn't understand it : it recognizes 8 or 16 bits only. For the buffer problem : have you tried to lower "Max blocksize" and "Min blocksize" ? I had the same problem, being quite frustated with a 10000ms buffer, telling everybody that the waveOut plug-in was really crap, until I set "Max blocksize" to 16kB and "Min blocksize" to 4kB. And now, all my problems have disappeared : I can even lower the buffer to 300ms ! Well, yes, I have a quite old system (PPro233) with a Marian Marc 2.
And like I even need to say this, MAD at 24-bit resolution blows Fraunhofer or audioEnlightenment out of the water. I'm using a Creek OBH-11 headphone amplifier mated to Sony MDR-V6 Studio Monitor headphones, and the difference is pretty stark and obvious. The MAD output sounds more natural, has better instrument decay, has more bass power, and the bass is smoother and the treble clearer. Fraunhofer sounds positively fake in comparison.
I agree completely. Even I don't have audiophile equipment (cheap Sony MDR-CD270) and my drums are not "audiophile" too (especially the right one, little problems when I was young), difference is clearly audible, when compared to NullSoft MPEG plug-in...
Bye,
Pierre Darbon
Pierre Darbon wrote:
For the oscilloscope (and VU meter), it's normal. MAD plug-in send 24 (32) bits and the visualization system doesn't understand it : it recognizes 8 or 16 bits only.
I wonder if it's possible to make this work in future versions of the plug-in? Possibly send 24 bit to waveOut and 16 bit to the visualization system? Or are the two one and the same?
For the buffer problem : have you tried to lower "Max blocksize" and "Min blocksize" ? I had the same problem, being quite frustated with a 10000ms buffer, telling everybody that the waveOut plug-in was really crap, until I set "Max blocksize" to 16kB and "Min blocksize" to 4kB. And now, all my problems have disappeared : I can even lower the buffer to 300ms ! Well, yes, I have a quite old system (PPro233) with a Marian Marc 2.
Thank you, that solved the problem. But there's no difference in using 24 bit and 32 bit resolution is there? One's just padded 24 bit. But since I can use 24 bit resolution successfully, I'll do a real hearing comparison. If anyone has already done this test, please share your conclusions.
Is there any GREAT difference between 24bit & 32bit? since 32bit is just padded 24bit if I'm correct), they should sound the same right?
About the vis, sounds like we need to make a plugin too or write to Nullsoft. I suggest we write to Nullsoft.......
& damn I really need to get a 24bit soundcard.
kaiwei
OK, I have some (but not a lot of) time to devote to MAD again. :-)
Re: 24-bit vs. 32-bit output
There should not be any difference in quality; they are identical except for the least significant byte of 32-bit output which is always zero. The only reason to have 32-bit output is that some hardware will have an easier time with 32-bit words since it is an "even" number (4 bytes instead of 3.)
Re: oscilliscope and VU meter, et al.
Pierre is right, it's a problem with Winamp that these visuals don't grok 24- or 32-bit output. I *could* work around this by feeding a lower resolution to the visuals, but it would cost additional CPU. Like kaiwei, my preference would be for Nullsoft to fix the problem at their end.
Re: high quality audio setup
Nice work... kind of ironic I don't currently benefit from 24-bit output myself. Maybe Santa will replace my SB16 this year. ;-)
Cheers, -rob
-
David Shin -
kaiwei -
Pierre Darbon -
Rob Leslie