Arduino digital chromatic tuner

See the writeup at http://deambulatorymatrix.blogspot.com/2010/11/digital-chromatic-guitar-tuner...

Category:

Science & Technology

Tags:

• arduino
• rbbb
• digital
• chromatic
• tuner
• yin
• pitch
• detection
• algorithm
• electronics

[MilwaukeeGuitar]

Can you check the link on the Yin paper or put up a cite? Thanks.

[raptorofaxys]

Hm, the link seems to work for me..? Youtube does appear to split it and insert a space on a line break, however, so it doesn't work if you just copy/paste it into your browser's URL bar. (I had to remove a space after the "c" in "cheveign" for it to work over here.)

At worst, just google "YIN, a fundamental frequency estimator for speech and music" or "2002_JASA_YIN.pdf", it's in the first page in both cases.

Let me know if that still doesn't work for you!

[blueduck577]

would there be any way to reduce the latency between picking a note and the midi output?

[raptorofaxys]

The pitch detection algorithm works by successive refinement; MIDI is the same as the regular tuner, but with fewer iterations. I tried to tweak the code to achieve the best compromise between latency and precision, but there wasn't much more I could do on the software front (without significant effort) to make the thing run faster. Hotspots are written in assembler and the rest of the compiler-generated code is pretty good. The next easiest gain would probably be on the hardware front.

[duckythescientist]

Wonderful, this has inspired me to add this functionality into (hopefully) a custom homemade guitar body. But I do have a question: could you use a Fast Fourier Transform (which I do not fully understand yet) or something of the sort to make it polyphonic?

[raptorofaxys]

The Yin paper touches on polyphony - a challenge in its own right..! An FFT would be a great starting point, but the Arduino's atmega168 doesn't really have the horsepower required to crunch the numbers (though I'd be happy to be proven wrong ;-) ).

One reason I chose Yin is because, unlike FFT, it is a time-domain approach; at the required precision, the implementation only requires about 350 bytes of RAM.  On the other hand, it's not amenable to polyphonic pitch detection.

[gaijintendo]

You said 'what's up' and didn't give me time to answer, which is very rude. I wasn't too offended, but others might be. Love the project, plan on attempting a simpler one myself. Cheers

[dodgedart74]

congrats on getting the job.

[slimjim8094]

Ducky, a FFT would work. People have done it on the Arduino (there's assembly libraries out there), but you want a fast one - or perhaps two, maybe use one like a dedicated DSP and have it spit a list of significant frequencies over serial.

I have an ATmega 328, which has twice the RAM (2kb).

I'm actually working on a similar project, but for a two-tone radio dispatch system. The average of peak approach seems like it may be more suited for my purposes.

[Disenyador]

Another one who wants a tutorial of how to make YIN pitch detection using Arduino!