Hacker News Reader

Comment by parkaboy | Hacker News Reader

This is awesome! One question, though: why not just work with straight audio and convert to midi with some basic* signal processing assuming a use case of instruments/sounds that aren't crazy synthesized. You could then also use this app for certain acoustic instruments as well and make this run all on mobile, potentially with an audio jack dongle/splitter if connecting to a digital instrument. Could also focus the audio processing on specific acoustic instruments out the gate and role out algorithms one at a time for each kind of instrument (e.g. piano vs saxophone)

View on HN

> though: why not just work with straight audio and convert to midi with some basic* signal processing assuming a use case of instruments/sounds that aren't crazy synthesized.

Why would you start with that? Particularly since some of the best "digital pianos" are actually just MIDI keyboard controllers (weighted and all that) and don't actually generate any audio themselves?

As someone with a professional background in consumer hardware (and music production) my first thought is always "how can I solve this problem without hardware" first. Don't get me wrong--hardware is fun--but manufacturing and selling consumer hardware is brutal from a business perspective (of course fun in its own masochistic way). With hardware, you're dealing with annoying cashflow issues unless you can easily build to order (money tied up in inventory or even the supplies), supply chain management, inventory management overhead, logistics and fulfillment overhead, additional regulatory overhead.

My understanding from OP's product is that there is an inherent assumption that this is targeted towards people dabbling on standalone digital pianos that also typically have built-in synthesis/sound modules.

Anyway, I'm just giving my reasons for why wouldn't you start with a software approach to answer your question. Putting all that aside, OP's solution is really cool and, independent of the decision to go custom hardware, it seems like a great product solution for the problem it's solving.

To not start with MIDI means starting with audio. That means a microphone, preamp, digital audio interface ... all of which are significantly more expensive than the essentially zero-cost DIN or USB MIDI port that this device requires.

This device is a hardware solution. Hardware solutions are very popular in the music technology space, because nobody really wants to be screwing around with their laptop or whatever. A software solution that did the same means a computer that is running with a mic, preamp and audio interface at all the same times as the person produces sound.

The product in TFA has no dependence on the presence (or absence) of synthesis in the keyboard unit.

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.

Also to clarify, yes: I understand most digital pianos have midi baked in. It's an easy signal to get out. I think this is a great product concept and there's nothing wrong with it--people will buy it! Just spitballing if hardware and all the baggage a hardware business model comes with is necessary.

This just wouldn't work as a phone app. The main selling point of this is always on recording. When the keyboard is on, it records, nothing for you to do.

Only way this works in software is if you dedicate a device to always record and audio processing doesn't provide isolation. Ambient sounds may get picked up or loud noises could drown out your playing.

This item works, because it is hardware. I also wonder how this would work connected to something like the sonuus G2M V3. Would be nice if the audio processing was solved by someone else.

Thats for V2!

Reply on news.ycombinator.com