Getting better sound in your dj sets (Tony Andrews interview)

Maximlee, post: 30538, member: 2165 wrote:
[FONT=Tahoma] if you are playing 320 mp3s… all the subtle harmonics are simply not there… [/FONT]

This is a wrong statement!

it is true however that the lesser the sound is processed by the DJ (so, by the software/hardware he uses) the more accurate the sound will be (therefore “better”). Especially with the auto-gain/limiter combo things can get nasty. Given the fact that 99% of all today’s tracks are compressed and limited like there’s no tomorrow, there really is no point in using an input gain (the gain in traktor). Most likely, you are gonna cause the track to go clipping, which will in turn force you to rely on the limiter of Traktor, which in turn, its not that great of a limiter.

My advice is to never use auto-gain, in fact never use Traktor’s gain – if you have an older track that sounds “lower” in volume, normalize it in an external wave editor. If it still sounds “lower” in volume, just decide if you are gonna play it or not – maybe, as a production it cannot “stand” in a mix with another track. Traktor’s limiter on the other hand is useful to catch the odd clippings here and there when you are mixing two or more tracks – but thats just about all it can do. When you are mixing more than 3 tracks, if you dont know how to mix them, the limiter will not suffice – the sound will go out nasty and distorted.

P.S: the reason I used “lower” in brackets is because sometimes a track’s perceived loudness is lower – that does not necessarily means it actually is lower in volume. Perceived loudness is dependent on stuff like the dynamics processing that has been done to the track (compression, limiter), the orchestration-genre, the structure of the mix and the mastering process – in these cases, gain cannot do much because usually the track is not actually lower in volume – it still reaches 0db so there is no point in applying gain. In fact when you do apply gain to it, you just cause it too get “extra-limited” which, with the Traktor’s limiter will not yield the best of results. Long story short, a track which is perceived lower in volume but is not, is a hard thing for a DJ to fix.

Maximlee, post: 30560, member: 2165 wrote:
Softcore you say that my statement is wrong and im not really in a postion were i have a system to prove it but you have to ask yourself the question… what is happening when you convert a wav file to mp3? you cant not argue that information is compressed and lost. there is 1091 kps taken out… you have took out rough 4/5 of the information and your saying that there is no harmonics and overtones in that.

Of course we have a degradation of quality because, of course, we have subtraction of audio data. What I meant is that the specific statement: “subtle harmonics are not there” is wrong. Do make a research on what actually science defines as harmonics and overtones and then you will understand that mp3 encoding is a lot more complicated than just saying that it takes harmonics/overtones out – the statement simply doesnt makes sense.

Secondly, I will disagree with the notion that our bodies feel a greater range of what science, up to now, tells us we are able to hear. Unless again, you have any official research and theory to provide me links to – its easy to all just go about and talk based on our thoughts, ideas, theories and/or misconceptions/preconceptions but when I discuss about sound in the way it should be discussed, scientfically, all these notions have to go out the window.
Surely, i’ll say that very low bass can be “felt” due to the vibrations it is causing to the environment and our own bodies – even then one would argue that its not the sound we feel, its the effects it has on our surroundings – you cant tell by the bass rumbling of a room anything about the tonality of the bass (what note is it)…As for higher frequencies? I higly doubt so, again if there is any official creditable research you can point me out to, I’ll happily study and learn more about my favourite subject: sound. 😉
The “i felt it with my body” thing, for me is definitely placebo effect when we discuss about sound quality. Finally, we didnt even touch the subject of WHAT exactly speakers one used to feel such frequencies as all the speakers I currently know of, are built on the notion that we do not listen to anything lower or higher, WHAT tracks we listened and felt such frequencies since all are made with content in the standardised 20hz-20khz range and generally what equipment one used because all the manuals of all the equipment of sound do tend to prove that nothing is reproduced lower than 20hz and higher than 20khz. So how does anyone feel anything?

128mp3s DO sound worse compared to .wav and that is not due to “pur bodies feeling the sound” – there is a huge difference in the reprodcution of especially the high frequencies where an effect similar to ring modulation takes place and the stereo image in a low quality mp3 format.

Finally, I will totally agree that, as you mentioned, we should strive to educate listeners to at least use more sensible formats and sound reproduction devices because thats where the real problem is. In that sense, this is why Im not such an enemy of the mp3 – its a format, which when used with a bit of knowledge, you can have a decent result to the end user without taking up big space in his storage device.

… I miss vinyl…. Beats any WAV/MP3/FLAC/whatever.

I do not miss vinyl a bit and I have DJed on vinyl for over 10 years…. (I think I still have like 10 unopened needle boxes in the basement, got to sell those too…)

Terry_42, post: 30675, member: 1843 wrote: Well what I did with a friend who is an audiophile and has this equipment:
Hook up my Digital Out (optical out) from my Computer to his Cambridge Audio DacMagic (yes this is the best DAC in the world).
From there into a Cambridge Audio Azur hooked up to a pair of Linn Akurate.
Test floor was a soundstudio of the ORF (Austrian Television and Radio station) with the listener perfectly in the middle.

Test Audio Files were 1x Apple Lossless Codec and 1x 256k AAC (to simulate iTunes).
Tested were: 1x Symhonie Fantastique by Berlioz performed by the London Symphonic Orchestra – taken from Japanese CD Version (as the Japanese CDs have the best quality audio).
1x Keith Jarret Tokio Live Concert (again Japanese Version)
1x Theme from Star Wars – Soundtrack CD – Japanese Version

Testers did not know what file was playing, file was switched (perfectly synced) every 15 seconds or 30 seconds.
4 Testers: 1 Audiohile, 2 fellow DJs and 1 studio technician.

None of the testers were able to distinguish the difference with 100% accuracy. The Audiophile Tester with the Berlioz track was the only one to get slightly better than “statistic guessing”. All others were within statistical margins aka did not know what was playing.

Now that is an interesting one. 🙂
So one tester was actually able to be significantly better than 50% right? (around 50% would mean pure guessing.)
How many times did you repeat the test? The more often you repeated it, the more meaningful we can consider the result.
Was the lossless source file in CD-Quality (44kHz, 16bit) or in higher resolution? And was the AAC file converted directly from the lossless one? If the source was a 24-bit file, could it be the software converted it back to 16 bit in the AAC conversion process (dithering etc. can also make a difference)? Have you taken care of the volume levels?
I’m not trying to nag about it, just trying to rule out anything that could have spoiled the result.
Also, in the end we have to remember that this is (like every blind test) only a comparison of a particular encoder of a particular audio standard with particular settings on a particular music sample. It could be better or worse if we change any of those variables.
If anyone wants to do any further testing, i still recommend doing it with foobar2000 because, as Mr. Andrews already pointed out in the interview, it’s known not to tinker with the audio unless you tell it to do so, and the semi-automation of the ABX-Plugin reduces some of the possibilities of human error that can occur when conducting a blind listening test.

I’m forced to A/B testing all the time here in the studio. Work in 24-bit mode and I dump things out as MP3 or AAC stubs. Sometimes I have to go back to the master and tweak it so the final MP3 does not sound so harsh. I think the dynamics are fine but I really miss the smooth vinyl sound myself concerning lossy codecs. Shows up with hihats and other cymbals, for example, or anything where the frequency range switches really fast and the lossy codecs halt along and try to do compromises each millisecond.

To do this kind of decent A/B test, you need to start with a good reference monitor setup. And streaming audio from the web with 128k will not cut it. You need 1160k WAV versus 320 or 192k MP3.

My guess — and it’s a guess — is that with 80% of the DJ setups and 90% of the consumer setups nobody will hear differences. However, with really good PA systems there’s always the weakest link…