Timeline for reconstruction filter - How does it actually work?
Current License: CC BY-SA 4.0
5 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Nov 6 at 3:09 | comment | added | 比尔盖子 | I guess storing and streaming high-resolution audio samples was probably a hack that once served an useful purpose in audio production in the era of NOS ADC/DAC. One would process and store high-resolution or oversampled data during production, before applying decimation or the ~20 kHz filter when the work is done. With modern DACs, the input is oversampled internally anyway, so this practice no longer makes any sense. | |
| Jul 29, 2019 at 14:14 | comment | added | Justme | Well realizable digital filters can't see into the future, so they are causal and work only with present and past samples. Therefore a symmetrical FIR filter with N taps has a delay of N/2 taps. And the filter has some sane amount of taps, only enough to have enough filtering within the specified tolerance. Or it could be an IIR filter. Or biquad. Or multiple cascaded filters. Or no filter at all, just add zeroes between actual samples and let the analog filter do the job. | |
| Jul 29, 2019 at 12:58 | comment | added | user1202136 | "Same thing when playing audio back, upsample and reconstruct digitally the signal to 192 kHz for playback": It is here that I struggle to understand. Say $y[4n]$ is your upsampled output at 192kHz and $x[n]$ is your 48kHz input. For computing $y[j]$ exactly, don't you need to know all $x[i], i=0,n$? If yes, how do you upsample in real-time, when you don't know input samples past $j/4$? | |
| Jul 29, 2019 at 12:17 | history | edited | Justme | CC BY-SA 4.0 |
Add reconstruction filter info
|
| Jul 29, 2019 at 11:57 | history | answered | Justme | CC BY-SA 4.0 |