I’m using one of the new Tsunami boards with the QWIC interface but I am controlling it via serial. I am controlling from a Teensy 4.1 and using the Arduino for coding which includes the Tsunami library. The Tsunami has the latest firmware - Tsunami2_20250621_v220s
The issue is that when I set a track to have a release time the sound during the release has a noise as it fades. It sort of reminds me of digital switching. As I increase the release time the noise almost seems to stretch out.
Tsumani is part of a bigger project containing several Patch SMs and the Teensy controller but I have taken the whole project down to just the Tsunami and the noise is still there. I do have an older Tsunami inside another project that I could drop out at a pinch but I’d rather not.
I have tried another uSD card as well with no improvement.
I suppose I have come here with the hope someone will say “yeah that’s unfortunately a problem the Tsunami has”.
In the meantime I continue to try things out.
My 1st guess is there might be a ground loop issue…What does the grounding situation look like? Are the teensy, audio speaker, and tsunami all powered independently?
Do you see the same thing with other test/demo tracks?
Use uncompressed 44.1 kHz, 16-bit stereo WAV files and see if it improves.
It’s a single power supply for the Teensy and Tsunami. I don’t think it’s a grounding issue as the sound is only there when the track is fading due to the release. I might try taking both the Teensy and the Tsunami out of the machine and just run them together.
The noise effect happens on most tracks. It sort of sounds like distortion but the level is certainly not overloading any part of the Tsunami that I can tell.
The files are definitely 44.1khz 16 bit stereo as a Tsunami can’t play any other type of file. And yes it originated as a WAV file and there is no compression degradation in the audio.
Is it a 32-voice or 25-Voice Tsunami Qwiic (Does the CPU end in N20 or N19)?
Can you provide the actual Arduino library commands (the code) that you’re using to play the track and start the release?
One thing that would be very valuable is if you could do the following:
Create a 30 second stereo file containing a 400Hz pure sine wave. Use the same code to play the sine wave and quickly implement the release. Record the output using a good audio interface and software such as Audacity. Post the results (just the release part.)
Thanks for the reply robertsonics.
The CPU is a N20.
I decided to isolate the code that I was using which still produced the same noise result:-
void loop()
{
// Call update on the Tsunami to keep the track playing status current.
tsunami.update();
// play track
tsunami.trackPlayPoly(1, 0, 0); // play track 1 via output 1 and leave the track unlocked
delay(2000);
tsunami.trackFade(1, -70, 2000, 1); // fade track 1 to -70db over 2000ms and stop the track when done
delay(5000);
}
Note, I did remove the tsunami.update() and got the same results.
I can’t seem to find a way of uploading a wav file so I’ve hosted it at my web site here - Amongst Myselves - Tsunami Noise Issue
For your reference I am using the Tsunami Arduino library with a date of 11/22/2016 in the header. I’ve not checked to see if this is the latest.
Steve.
When fading up or down, Tsunami steps through a dB lookup table to so that the fades are linear in dB. This table is in increments of 0.5dB which is the gain granularity. When a gain change is rquired for the fade, it’s ramped on a per sample basis over one audio buffer so that there are no discontinuities. But if you are fading over a long time, then that gain will remain constant for a number of audio buffers, until the next 0.5dB change.
I think what you’re hearing are just these steps. If you examine the sine wav file, you can see them. I’ll compare these results to the WAV Trigger and WAV Trigger Pro to see if there might be some issue causing more stair-stepping than intended on Tsunami, but this will take me a few days. I’ll follow up on this thread.
Thanks robertsonics, I presume it’s Jamie behind that avatar.
It’s one of those issues when I’m making my project and hear this sort of thing and want to find the origin. It’s probably not an issue for me but just a case of finding where the problem has stemmed from. I might pull out my older Tsunami from my FloriVoxTron project to see what that does.
All of the Robertsonics players theoretically use the same fade algorithm with the same 0.5dB lookup table, and the same algorithm is used in a commercial audio product, so it should be acceptable. However, it’s entirely possible that the latest Tsunami firmware is producing different results for some reason, so I just need to confirm one way or the other.
I’ve just pulled the Tsunami from a previous project and the fade artefacts are not there. This board is the pre-Qwiic version. N20 processor. That has the AK4816VQ DAC. I’ve also placed this on my website - https://amongstmyselves.com/tsunami-noise-issue The file is the preqwiic-tsunami-fade.wav