Yes but, an M7 is not 48 bits in or out. A typical input signal will use more like 16 bits out of the available 24. That input signal is then multiplied/divided, EQed & compressed Ö more mathematics, assigned to a VCA and multiplied again, added to a bunch of other signals. The signal path is changed to 48 bits at some point, the signal is then modified by a GEQ and compressor and scaled back to 24 bits.If you lose a little bit each time it may well become audible, add to that less than perfect AD & DAís and some ordinary Mic preampsÖ. it all adds up.
I have AB-ed an M7 directly with an iLive and an iLive with a Pro 2. The Midas sound great, the iLive also sounds very very good, the M7 is awful, and to my ear the M7 seems to get worse on more complicated mixes.So Iím sticking with my Ö I wasnít there comment
So after reading this I would conclude that my X32... With the fine sounding preamps that I have verified with many multitrack recordings, out thought the AES digital out... Should be a better sounding console than an LS9...all other things being equal.
What exactly does it add up to? I am an old analog dog who would love to reveal some tragic flaw in digital audio, but just don't see the smoking gun. Your same logic applies to analog consoles, where every gain stage or EQ circuit the analog signal goes through "adds up" and degrades that signal. In fact a simple digital multiply or sum of multiple stems results in more bits of resolution so arguably lossless, certainly less deterioration than an analog path. There is always room for better or worse execution of things like sample rate conversion, dynamics, and especially effects, not to mention the myriad ergonomic factors associated with consoles that affect perception of performance. People have been sniffing around this vague complaint about digital audio with no smoking gun revealed yet AFAIK, and I have been following this for a while. I never argue with people about what they say they hear, while I don't have to agree with their conclusions. JR
I think the same logic applies to analogue, and there are good and bad analogue desks, the more signal processing you do the worse it gets, but in general I find bad analogue easier to listen to than bad digital Örelatively speaking.
When pocket calculators first came out I can remember testing their accuracy by taking the sine and arcsine of a small angle. You never got the same angle back, some calculators got very close, some did not.
I was talking to a friend the other day that has a good understanding of the internal workings of DSPs and audio. He seemed to suggest itís the same with DSP processing and in some cases the designers take a few mathematical short cuts. 01 + 01 does not always equal 10.
Sooo Ö I think it is worthwhile looking at these types of issues and not just blaming everything on the sound of the mic pres as everyone seems to do.
I donít know how much the internal DSP processing or the quality of the AD & DAs contribute, but I think itís worthwhile doing some more complex testing. Running a 16 bit CD through a line-in or saying 1 Ö2 through a single mic is not enough to determine how good the audio quality is. Thatís not to mention using an MP3 tack from an iPhone.
I remember when MP3s first came out we did some AB testing against CD and DAT for an AES meeting. On simple vocal tracks MP3 sounded fine to me, but on the more complex tracks to my ear they became gritty and annoying to listen to. It was hard to pick out the individual instruments in the mix. This is how I find the M7, but with most of the new digital desks I struggle to hear any difference. Peter
...design of M7 is a throwback IMO.
Just for reference I'll throw this out there... thinking back to my days working on Data General ALU cards, multiplies are basically shift lefts of a binary word, an example:01 shift left 1 bit (multiply x2) = 10What gets filled in on the right?
Zeros, in a pure computer number, but do DSP designers fill with something approximated, or random/dithering?10101010 x 4 (shift left 3 times) = 1010101000Division goes the other way, shift right.
Page created in 0.156 seconds with 24 queries.