tautologico said:
I didn't say it is impossible, only that it's hard.
And I did not say you said it was impossible. I said it wasn't hard at all.
tautologico said:
It's certainly not as simple as a tuner. You mentioned Guitar Rising, a vaporware that had only a concept video on the site, so it does not count as proof.
It is, actually. Such a mechanism does not need to make complex analyses of input notes. All that is required for a real guitar to be used as a controller for a note-matching game is for two wavelengths to be compared and matched. Just like a tuner does. The game analyses the note you input from the guitar, compares it to the note speeding past on the screen (or, rather, the note behind the graphical representation on the screen), and if the wavelength is similar enough (threshold would be a big part of getting this right), and matches in time (rhythm), you score a correctly struck note. Since Guitar Hero/Rock Band songs are re-recorded, comparing single notes of any kind of instrument is eminently easy.
I'm fairly certain you are not going to believe me, but the sound input engine of Guitar Rising was very real. I've seen it demoed, in real life.
Generic question: If it is so easy, why not before? Two good ways to approach that: 1) It has been done. There have been several tech demoes of this very thing, some of them working with almost perfect accuracy. 2) The market is believed to simply not be big enough to warrant an AAA title with this mechanism. This analysis might be wrong, but it sure does influence devs over the globe.
tautologico said:
I know one or two things about signal processing, and I mentioned the possibility of processing the signal in my post. But one thing is processing audio signals as input, to output audio signals (with added effects or whatever), while another thing is processing the analog input signal to output a simple, digitized, midi-like output. Possible, yes, but not easy. For example, see what some guys that do this for a living have to say here [http://www.akoff.com/about.html]:
What is Music Recognition?
In a few words music recognition is mathematical analysis of an audio signal (usually in WAV format) and its conversion into musical notation (usually in MIDI format). This is a very hard artificial intelligence problem. For comparison, the problem of recognition of scanned text (OCR - Optical Character Recognition) is solved with 95% accuracy - it is an average exactitude of recognition of the programs of the given class. The programs of speech recognition already work with 70-80% accuracy, whereas the systems of music recognition work with 60-70% accuracy but only for a single voice melody (one note at a time). For polyphonic music the accuracy is even lower.
Yes, I could tell form your first post that you applied real-world knowledge. And your analysis of the situation you describe (as compared to what I speak of) is spot-on. I don't doubt your knowledge.
But you approach the implementation wrong. Just like the piece on Music Recognition above does not pertain to this problem/solution. What you and they speak of, is the artificial intelligence required to make complex analyses of wavelengths, and to output notation. A music matching game simply needs to match wavelengths, as per my explanation above.
tautologico said:
Well, 60-70% accuracy for a single voice (the guitar, in this case) is not good enough for a video game, I think. Also, if it was so easy, Harmonix could just support this in RB3 (that already has a Pro mode) and there would be no need to create a special guitar/controller hybrid. Yes, it is a niche, but for Harmonix, they already have the game, it would actually open up a new market.
The special controller/hybrid constitutes a large part of the income of guitar rhythm games. It is not created out of need, but out of want. And yes, Harmonix could support this in RB3 easily. But there are MANY reasons software does not implement functions, not only the difficulty of doing so. The whole concept of feature bloat is based on this, as is the idea of releasing several games with different focus, instead of cramming it all into one. The simple fact that Rock Band and Guitar Hero are games, intended for casual entertainment and party consumption, has been reiterated by the devs of those (and other) music games time and time again.
The interest of rhythm games developers in gradually providing more complex control methods for their music games is evident in the addition of Pro mode for Rock Band 3, something the Rock Band devs recently explained was a "leap of faith", but gradual is an important word here. It is entirely possible we'll see real guitar input for an upcoming Rock Band/Guitar Hero title, but the fact that it is not here yet, is a result of an uncertain market that is still very young (it might be easy to see the music game market as saturated, but compared to most other genres, it is positively undernourished, and only in its infancy), hard to analyse, and unpredictable. Not any technical limitations.
tautologico said:
processing the analog input signal to output a simple, digitized, midi-like output
This is the basis of drum triggering, and of using soft synths as sound replacement, and is done frequently, rather easily, and often live.
In the end, the fact that I've seen this done is enough for me, and voids the need for the explanations I give above. But unfortunately, others rarely trust such statements online.
EDIT: Hell, all song input mechanisms (Singstar, Rock Band, Guitar Hero) compare pitch. It's a bloody industry standard :-D A real guitar as a controller would work just the same: check the pitch/wavelength of the input, and compare it to the required pitch. Vocals, guitar note, strange bleeping noises, it makes no difference. It is all simple, and done on a common basis.