gregorio
Headphoneus Supremus
- Joined
- Feb 14, 2008
- Posts
- 6,746
- Likes
- 4,035
[1] Isn't time of an audio signal the same thing as its frequency? ... the point I'm trying to make is that if the frequency response is linear by +/- 0.5 db, then timing must also be perfect within the +/- 0.5 db range across the frequency spectrum - or am I missing something here?
[2] ... some claim that jitter and timing errors can be an issue with DACs (and of course this leads to all sorts of marketing claims to support their $$$$ DACs),
[2a] but how do those claims sit with, for example, a CD which measures linear in frequency response at +/- 0.5 db?
1. Yes but we're all talking about a number of somewhat different things here and lumping them together as the same thing. For example, at the simplest level, if we have a sine wave at say 1000Hz at 80dB (peak or RMS) and change it's time, say slow it down by 1%, then it will no longer be 1000 cycles per second (cps, 1000Hz), it will be 990cps/Hz. In this example the amount of amplitude (dB) is completely unchanged, it will still be 80dB (peak or RMS) just the pitch/frequency changes. However, when were dealing with more than one instance of a signal (even the exact same signal) then we'll have a summing/cancellation effect. For example, take our sine wave again but now add an identical one that's delayed by 1 milli-sec: As it's 1000cps, that's 1 cycle in a milli-sec, therefore the second instance is 360deg out of phase, sum them together and the result will be double the amplitude (86dB). However, if the delay is half a milli-sec then the two instances will be 180deg out of phase and the result will be complete cancellation (the amplitude will be 0dB). So, depending on the time delay between the two instances we will have double the amplitude, no amplitude or anywhere in between. By the same token, if we take the same delay, it will have a different effect on a different frequency. So our 1 milli-sec delay will cause a doubling of the amplitude of two 1000Hz sine waves but will completely cancel out two 500Hz sine waves (as 1ms is half the cycle time of a 500Hz sine wave and it will therefore be exactly 180deg out of phase). Consequently, if a sound contains harmonics at 500Hz and 1000Hz and we have a second instance of that sound that's delayed by 1ms, the amplitude of those harmonics are going to change drastically (the 1000Hz harmonic will double and the 500Hz harmonic will disappear) and therefore the timbre will be different (as "timbre" is defined by the number and balance of the harmonics in a sound). This is all on paper though, in practice we won't just have two instances of the sound, we'll have two signals (stereo speakers), numerous different instruments/sounds and notes, each with numerous harmonics, numerous different room reflections and each with a different time delay. So it's all a bit of a mess and we rarely get perfect doubling or cancellation, we just get some amount of amplitude alterations of the fundamentals and harmonics. To further complicate matters, how we perceive these alterations in the amplitudes/balance of harmonics varies considerably, depending on context, expectation and other factors.
2. Jitter is an entirely different type of timing error. With jitter we're not talking about timing differences between different instances of a sound/s that result in summing/cancellation "phase" effects but if we were, then as jitter is in the pico-sec rather than milli or micro sec range, we'd be talking about significant summing/cancellation of signals with wavelengths above the giga-Hertz range! Jitter can be "thought of" similarly to quantisation errors, we effectively get a perfect signal reconstruction plus some distortion product, a "jitter spectrum" which is effectively random noise with the odd peak here and there. A decent (sub $100) DAC/player should reduce the totality of jitter artefacts to below -120dB.
2a. It's effectively part of that "linear frequency response within +/- 0.5dB". That "+/- 0.5dB" is the total amount of frequency distortion, a combination of all the distortions in the analogue components of the DAC/player and the digital signal processing and reconstruction (which includes oversampling, reconstruction filter and jitter artefacts).
[1] OK. I was saying that you aren't liable to find timing errors of the millisecond range in home audio electronics.
[2] We just talked past each other there.
1. Yes you will, because it's likely that the recording the home audio electronics are trying to reproduce will contain timing errors in the milli-sec range and that's ignoring any reflections of the listening room/environment.
2. Yes, I agree. You seem to be talking purely about the timing errors introduced by the "home audio electronics" themselves, while I'm talking about timing errors which are going to exist anyway, regardless of the home audio electronics.
G