From CCRMA Wiki
by Uri Nieto
Idea / Premise
Game like Guitar Hero but for singing/growling mainly focused on metal music.
The singer/growler/screamer will follow the song being played, and Growl Hero will analyze it and tell him/her how good he/she is. Feature extraction of the signal would be necessary in order to achieve this. The pitch detection will also count, so in the end two different aspects of the voice will get involve in the game: the pitch and the type of scream.
Screaming has always been a part of my life. I’ve been in a metal band for the past 9 years, and I have been researching on different types of scream. I love Rock Star, but I think it lacks the real screaming part. This project would add this feature making it available for this extended type of music.
- what is the final product?
It is a game where one can scream/sing/growl and the game evaluates you and gives a punctuation. In the end users could add different songs if they define the pitch and the of scream for each sample of the song.
- what is the interface?
A screen where your signal is showed and it should match the pitch and the type of growl of the song being reproduced. what is the user experience like? Singing and Screaming and Growling until they get the higher punctuation. It would also be nice to have a training mode.
- what is the aesthetic of the software?
Apart from the signals, it could have nice graphic as a background. Something devilish probably…
You will hear the music and sing/growl/scream following it (theoretically, you should have a mic further away from the sound source, so that the input doesn’t get the sound coming from the output). There will be marks on the screen following a scrolling time signal where the Pitch, the lyrics, and the type of the scream will be available at all the times. User will be able to pick up the song.
- real-time interaction? Yes
- graphical / textual, both? Both
- networked? No (maybe for future work)
- software design / architecture: I will use OpenGL for the graphics, RtAudio for the input/output audio and XCode for programming an app (only for Mac for the first release).
- how/who will test it?
Basically me, but I’ll let any other CCRMAlity try it out, and actually anybody that might be interested. what is the measure of goodness? I will have to pick up a nice tolerance level to decide whether the scream is good or not. Probably I will have to calibrate to program at the beginning of the execution to decide what level of scream is good or not.
- 1: Efficient Pitch detection real time algorithm
- 2: Feature Extraction of the different types of Screams
- 3: DESIGN!
- 4: Sync input sound with output music and evaluate the growler/screaming/singer in nice GUI
- 5: Combos for punctuation, World Ranking, Training Mode,…