Optimizing for the Daisy seed was definitely the challenging part! But honestly it’s such an easy platform to develop on. The folks at Electrosmith have done a great job making prototyping simple.
Really good job! Sounds fantastic. How does it sound with a boost in front? Would be great to hear this at the end of a pedalboard to get an idea of the headroom it has!
I need to do more testing with that, it sounds OK with a pedal before it, but the Terrarium pcb is powering the buffer circuit with only 5v (most pedals use 9v, some even 18v). In most cases, using the input gain knob (digital gain) on the pedal will likely sound better than boosting the actual signal prior to the pedal.
@@GuitarML Makes sense. Also, since the WM8731 runs at 3V3 on the Daisy, there's only so much the op-amps can squash the signal to before the noise floor starts creeping up. But I guess that's the price you pay for digital processing of any kind.
That’s what I originally wanted to look into, but I think NAM just takes too much processing to run on the Daisy seed. The models I’m running here are significantly smaller
Jeez, it's mindblowing. Thanks for the great work and sharing this knowledge. Amazing
Excellent work! Thanks for releasing the source code with it, too. Always something to learn from others.
Incredible amount of features & performance for such a fairly low-spec embedded platform.
Definitely going to check out the code.
Optimizing for the Daisy seed was definitely the challenging part! But honestly it’s such an easy platform to develop on. The folks at Electrosmith have done a great job making prototyping simple.
So cool - thanks for open sourcing the code and the process. The 57 sounds pretty good ... you've basically made your own UA Woodrow clone ;-)
epic project, thanks for the code.
That's really impressive tbh, bravo !
So cool!
This is lit!
Love this idea! The models inference is running right on the daily seed mcu?
Really good job! Sounds fantastic. How does it sound with a boost in front? Would be great to hear this at the end of a pedalboard to get an idea of the headroom it has!
I need to do more testing with that, it sounds OK with a pedal before it, but the Terrarium pcb is powering the buffer circuit with only 5v (most pedals use 9v, some even 18v). In most cases, using the input gain knob (digital gain) on the pedal will likely sound better than boosting the actual signal prior to the pedal.
@@GuitarML Makes sense. Also, since the WM8731 runs at 3V3 on the Daisy, there's only so much the op-amps can squash the signal to before the noise floor starts creeping up. But I guess that's the price you pay for digital processing of any kind.
I want to get in to that. Looks like fun. What was the IR you used??
It’s the one I used in the Proteus plugin, combined fromseveral other IRs: github.com/GuitarML/Proteus/blob/main/resources/default_ir.wav
Is that possible to run NAM files on it?
That’s what I originally wanted to look into, but I think NAM just takes too much processing to run on the Daisy seed. The models I’m running here are significantly smaller
Impressive, how is the latency ?
I hadn’t measured until you asked, it’s just over 2ms!