Sorry I'm not sure you're following me. I'm suggesting that the user's phone that they already need (since the hardware pairs with the app in the article) is t

Sorry I'm not sure you're following me. I'm suggesting that the user's phone that they already need (since the hardware pairs with the app in the article) is the hardware in this context. A phone already has a microphone and--if the environment is too noisy- many also support direct audio input through the phone's charge/data port with a simple off-the-shelf converter. Most smartphones can perform real-time audio progressing nowadays. The trade-off is that midi is going to be lower error than an audio processing algorithm and much easier to implement out-the-gate, while it's certainly a step-up in the software engineering needed to develop and implement an audio processing algorithm to implement pitch/key detection as well as capture relative timing and velocity. Correct the product in the article has no dependence on synthesis, but in pretty much all contexts in which someone will be using the device, there will be a corresponding audio signal present. Most people don't tap out music with a midi controller just in silence/without some sort of synthesis in conjunction.