Is this a silly question?

[Arizona Dan]

I have been wondering why music played via loundspeakers sound better when played with a live feed directly from a microphone than when played from a recorded source - even 24 bit high resolution.

What I am really wondering is What would it take to design/build an A/D and D/A that would sound more like a direct feed. Just as an experiment. Forget about being practical. Even if it took 1 Tb to record a song, or fraction of a song. Here's what I am thinking. Make the D/A as simple as possible. Remove all filters, algorithims, etc. Starting at the beginning, the microphone converts air pressure flucuations (music) to voltage. Record the voltage from the microphone at a very high sample rate (whatever it takes for this experiment) and at high precision (again, what ever it takes). For playback, reverse the process. Convert the data back to voltage. No filters, no algorithims, just reverse the process as simply as possible. Make the sampling rate high enough (again, what ever it takes) that we do not need to guess what is happening between data points. Then feed the signal through our stereo preamp, amp, and then to loudspeakers.

Does any one know how to perform such an experiment? Just for fun.

Dan

[jimbo1968]

Originally Posted by Arizona Dan

I have been wondering why music played via loundspeakers sound better when played with a live feed directly from a microphone than when played from a recorded source - even 24 bit high resolution.

Does it ?

I imagine having the performer live in front of you could skew your perception quite a bit.

[TheSeekerr]

Yeah, I'm leaning towards "silly question" too - a lot of pro audio is done via digital mixers these days, which are an ADC>DSP>DAC system - no-one I know thinks they sound worse. I suspect you just have an unconscious live performance bias.

[Nrik]

No - there is a real difference: Dynamics!

All recordings (yes all!) use dynamic compression. That is the reason.
Ask a local studio-technician if he could show you how he works one day, and a whole new world appear for you.
A lot of instruments actually sounds better with some compression, and he ofcourse also have to watch the limited dynamic range of a 16bit CD (96dB).
...and then there is the whole "loudness war" thing.

Even with 24bit sources (which has more than 130dB dynamic range) technicians seldomly use the extra headroom to the max. Probably both because of clipping risk and 'perceived' sound quality.

[Pano]

I used to do the "live vs recorded" experiment a lot. Live feed always sounded better. Why? Same gear, same speakers, everything.

Then one day I realized. The recordings were missing the direct sound from the stage. Doh! Even tho the direct sound was low compared to the amplified, it still added a nice realism.

Only did the isolated musician test once. Live feed still won but by a much smaller margin.

[TheSeekerr]

What I was alluding to with the digital mixer comment is that there's an easier way to do the comparison: compare a direct live feed to a live feed with an ADC>DAC in the middle of it. Personally, I'm willing to bet that in a blinded, level matched test, very very few people would pick the difference.

[Pano]

Well when I was doing the tests only the recording was digital (DAT). But I'm old.

[Arizona Dan]

Thanks to everyone for their replies. The February edition of Stereophile, page 3, touches on this topic. I really don't have any direct experience with digital recordings, but it would seam that the live feed has an advantage because the music is presented whole. The digital version takes the music and breaks it into pieces, filters it, tries to guess what data is missing between the data points, and then tries to put it back together again in an analog form. As good as high resoulution digitial is, the music is affected by the process.

If any of you are D/A experts, is it possible to eliminate all of the filters and algorithims and just create a voltage for each data point?

Dan

[TheSeekerr]

That's basically what a non-oversampling dac with no output filter does - produces a "stepped" output corresponding to the samples. But since you're now playing back a square-ish wave, you're introducing higher harmonics (which your loudspeakers probably won't reproduce, but some people worry about it).

[Arizona Dan]

Could you explain why it would introduce higher harmonics?

The second question is: If we live with the stepped output, how small does the step have to be to be inaudible? How would one decide? Just out of curosity, what is the sampling rate of HDTV? Audio is 16 bit-44.1 KHz up to 24 bit-192KHz. What is video? Is there any corelation between the sensitivity of the human eye compared to the human ear?