Topic: wtf?? record at 16 bit > dither to 24 then dither back to 16 bit flac??

[Gordon]

am I missing something here??

Source:
FOB (Neumann KM184 mounted 12 feet up at the front
left corner of the SBD console; row N, seat 107) >
Lunatec V2 > Benchmark ADK2+ > DAT(Tascam DA-P1, 16
bit/44.1 KHz).

Transfer:
DAT(playback on Tascam DA-45HR) > MOTU 896 > Apple G5
(Digital Performer 4.5 recording as 24 bit/44.1 KHz
AIFF > Waves L3 Multimaximizer -flat freq. response
curve, threshold: - 4 dB, ceiling -0.1 dB)> 24
bit/44.1 KHz WAV > P4 PC - Wavelab 5.0 (24 > 16 bit
WAV using Waves IDR mastering plugin set for 'Hi Res
CD Master' (Dither Type I, Ultra shaping) > CDWave
v1.71 (tracking) > FLAC (FLAC frontend).

[trajhip2000]

just a minor point, you don't "dither" the wordlength up, only down. but to answer your question, it's generally a good idea to carry out any digital processing at a higher wordlength (e.g. 24-bit or greater) and dither back down, it reduces the impact of roundoff error (distortion) if the math is done at a higher wordlength.

Steve

[F.O.Bean]

just a minor point, you don't "dither" the wordlength up, only down. but to answer your question, it's generally a good idea to carry out any digital processing at a higher wordlength (e.g. 24-bit or greater) and dither back down, it reduces the impact of roundoff error (distortion) if the math is done at a higher wordlength.

Steve

isnt that what temp files are for?

[dmonterisi]

i think you're right bean...i'm not sure what benefit you would get by recording a 16 bit signal @ 24 bit as those 8 bits would presumably be zeros, i don't think there is any up-res happening so i'm not sure what the point is either.

[nic]

actually, the transfer from dat is via analog to 24/44.1.

if he hadnt run the IDR mastering plugin, I would like to hear a comp of the original dat and resulting wav...

it's generally a good idea to carry out any digital processing at a higher wordlength (e.g. 24-bit or greater) and dither back down, it reduces the impact of roundoff error (distortion) if the math is done at a higher wordlength.

DP and most other audio apps work with files in a floating 24bit(or 32bit) format, so that is being done anyways...

[dmonterisi]

actually, the transfer from dat is via analog to 24/44.1.

i see...doesn't that make even less sense though?  wouldn't it be better to actually run an up-res/upsampling algorithm on the 16bit wav without involving the Tascam DA45-HR DAC?  it's probably a good DAC, but still...

[mmedley.]

just a minor point, you don't "dither" the wordlength up, only down. but to answer your question, it's generally a good idea to carry out any digital processing at a higher wordlength (e.g. 24-bit or greater) and dither back down, it reduces the impact of roundoff error (distortion) if the math is done at a higher wordlength.

Steve

If you are using 24bit or 32bit float temp files then yes it is basically the same thing. While you are not adding anything to the recording, you are 'splicing' the waveform into smaller chunks, that way when it rounds off it is more accurate as trajhip said. Think of calculus and integrals. Nice job Bean!!!