So I have been playing around with delays recently as I learn my way through all the pieces that are DSP. I remember actually going through the 1st edition book for FX as well as currently.
I noticed that even if the delay is set to say 5000ms, the delay is far far shorter in practise, 2000-3500ms. This may or may not be an analytical issue however, even in a daw such as ABLETON with the ms set at 5000ms, or 5seconds as it is displayed, the delay actually fires around the 3000 ms mark which even though it was longer than r7, it was still far off the 5 seconds.
I figure this is likely a compound error which is amplified as the ms time gets longer, somewhat how errors in responses can occur in filters that approach certain frequencies (high, low).
I was wondering if anyone has noticed this and could shed some light on this topic or point me to some resources that discuss this!Â
Could you tell me which project or object you are using for testing? Is this the CircularBuffer object or something else?Â
I will do some testing on the circular buffer object in the meantime.
EDIT: I tested the StereoDelay project that uses the CircularBuffer object and verified that with a 1000mSec delay time, each delayed signal was right at 1000mSec.Â
I used the Audio/MIDI -> Process Into WAV File with a single drum hit every 2 seconds, 0.0% feedback, and the 1 sec of delay and the resulting WAV file showed 1 second of delay between the 2-second hits.
I processed these for about 170 seconds with no drift in the delay time between hits.Â
Usually when my students report that a delay is happening at 1/2 the time (2500mSec for your 5000mSec delay) it is because they are not using a separate delay line for the left and right channels.
In one case it was because the student tried to interleave the circular buffer for L/R/L/R... but was skipping over one channel, getting only 1/2 the delay.
In another case using a circular buffer as a wavetable oscillator, there was a disparity because the sample rate was 48kHz and not 44.1kHz, even though the Windows driver was reporting 44.1kHz. It turned out to be a virtual box problem, and was not enough to cause a 1/2 pitch error, but it did cause a problem with a student with perfect pitch, who knew that the note was significantly out of tune.
When creating the circular delay buffers, stereoDelay.createDelayBuffers(resetInfo.sampleRate, 2000.0), I forgot to change the hardcoded max buffer value from 2000. So it is wrapping back around. In the end it WAS my own oversight.Â Might be a good lesson for someone in the future. Sigh.
Sorry for wasting your time with this thread.Â
So I wanted to test the plugin out under your 1000ms. I found this works perfectly. So I decided to try out intervals, 1000, 2000, 3000, 4000, 5000.
INTERESTINGLY...... 1 and 2 seconds were perfect. 3 wraps to basically 0 or more suspiciously 2.5ms.... (3ms maybe). then 4secs goes back to 1 second delay, and 5 takes on the 2s delay position.Â
SO after the 2000ms mark, something goes off. I am going to look through the code to see if I can see anything but this may go deeper! Id be interested to see if yours also does this.Â
Obviously a delay time that long isnt used that often, I only came across this while using extreme parameters to ensure my delays were operating as desired. When I saw this I was worried that my small delay times were off but it appears not which is fantastic actually.Â
Downloaded a fresh StereoDelay Project (with presets version);
Modified defaults as shown in picture (5s, no feedback, wet signall only)
Created WAVs from Audio/Midi to Wav file
Create WAVs from Bitwig Delay using an audio file same as inside R7 and from using the compiled SD plugin VST3.
Both appear to delay around 2000ms as opposed to 5000ms indicated in the DAW Export wav, with the corrected sample rate delaying slightly later.Â
I will try to re export out of ABLETON as well instead of BITWIG as I remember finding it produced a similarly short result in its delay but maybe I mixed up the signals. It is crashing today for some reason so I will update the thread with the findings of that when I get it working.
Most Users Ever Online: 152
Currently Browsing this Page:
Guest Posters: 1
Moderators: W Pirkle: 689