Skip to content

Commit 80b4fb0

Browse files
committed
added gonzalo's corrections
1 parent c44f3be commit 80b4fb0

File tree

1 file changed

+4
-4
lines changed

1 file changed

+4
-4
lines changed

README.md

Lines changed: 4 additions & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -5,11 +5,11 @@ Wekitunenator 3000 is our final project for the Advanced Interface Design course
55

66
[![Wekitunenator 3000](http://img.youtube.com/vi/M6bAb42s-lo/0.jpg)](http://www.youtube.com/watch?v=M6bAb42s-lo "Wekitunenator 3000")
77

8-
Wekitunenator is an instrument that applies sound effects in real-time to the user's voice. The applied sound effects are selected with a MIDI controller and the parameters regulating them are modified with the user's hand movements. Machine learning is used in mapping hand movements to sound effects values. However, this instrument tries to move away from the conventional notion of machine learning (ML) as a control paradigm and intends to use ML as a discovery tool. This was inspired by Sonami and Fiebrink's paper [Reflections on Eight Years of Instrument Creation with Machine Learning](https://www.nime.org/proceedings/2020/nime2020_paper45.pdf), presented in NIME 2020.
8+
Wekitunenator is an instrument that applies sound effects in real-time to the user's voice. These effects are selected with a MIDI controller and the parameters regulating them are modified with the user's hand movements. Machine learning is used in mapping hand movements to sound effects values. However, this instrument tries to move away from the conventional notion of machine learning (ML) as a control paradigm and intends to use ML as a discovery tool. This was inspired by Sonami and Fiebrink's paper [Reflections on Eight Years of Instrument Creation with Machine Learning](https://www.nime.org/proceedings/2020/nime2020_paper45.pdf), presented in NIME 2020.
99

10-
For modeling the mappings between hand positions and effects parameters, Wekinator is used. The model can be trained directly using the MIDI controller by selecting the effects (with buttons) and the parameters values (with knobs) and hitting on the `REC` button. The `REC`, `TRAIN` and `RUN` buttons control the MIDI interface (if no MIDI controller is available, the buttons in the screen interface can also be controlled with the mouse). This way, the user can dynamically train and use the model without having to use the keyboard or mouse. When running the model on unseen combinations of effects, the mappings and hence, the resulting sounds, will be unpredictable.
10+
In order to model the mappings between hand positions and effects parameters, Wekinator is used. The model can be trained directly using the MIDI controller by selecting the effects (with buttons) and the parameters values (with knobs) and hitting on the `REC` button. The `REC`, `TRAIN` and `RUN` buttons control the MIDI interface (if no MIDI controller is available, the buttons in the screen interface can also be controlled with the mouse). This way, the user can dynamically train and use the model without having to use the keyboard or mouse. When running the model on unseen combinations of effects, the mappings and hence, the resulting sounds, will be unpredictable.
1111

12-
The hand positions are tracked by a lightweight version of [@faiip](https://github.com/faaip/)'s [handPose-OSC](https://github.com/faaip/HandPose-OSC). The code can be found [here](https://github.com/gonski/HandPose-OSC). The communication between this model, the MIDI controller, Wekinator, and the Digital Audio Workstation (Reaper) is handled by the Processing sketch `handleOSC`.
12+
The hand positions are tracked by a lightweight version of [@faiip](https://github.com/faaip/)'s [handPose-OSC](https://github.com/faaip/HandPose-OSC). The code can be found [here](https://github.com/gonski/HandPose-OSC). The communication between this model, the MIDI controller, Wekinator, and the Digital Audio Workstation (Reaper) is handled by the Processing sketch `handlerOSC`.
1313
## Pipeline
1414
<div style="display:block;text-align:center"><img src="./.bin/pipeline.jpeg"/></div>
1515

@@ -28,7 +28,7 @@ Run `handlerOSC.app`. If HandPose-OSC is also running, you will be able to see t
2828

2929
<div style="text-align:center"><img src="./.bin/handlerOSC.png"/></div>
3030

31-
Select your MIDI input device and press `UPDATE`. The FX switches are controlled by MIDI pitch indexes 0, 1, 2, 3 and 4. The `REC`, `TRAIN` and `RUN` buttons can be modified with MIDI pitch indexes 5, 6 and 7. If you don't have a MIDI controller device, you can also control Wekitunenator clicking on the dark blue buttons. If you want to be able to modify the effects parameters during training using knobs in your controller, you will probably need to change the values of `knobCCs`in `handlerOSC.pde` and assign them to the corresponding ones from your controller.
31+
Select your MIDI input device and press `UPDATE`. The FX switches are controlled by MIDI pitch indexes 0, 1, 2, 3 and 4. The `REC`, `TRAIN` and `RUN` buttons can be modified with MIDI pitch indexes 5, 6 and 7. If you don't have a MIDI controller device, you can also control Wekitunenator by clicking on the dark blue buttons. The eight effect parameters are controlled by MIDI CC messages 48, 49, 50, 51, 52, 53, 54 and 55. You can also edit them within Reaper's interface with the mouse.
3232

3333
### Wekinator
3434
You can download Wekinator [here](http://www.wekinator.org/downloads/). Once downloaded, open the program and select `File>Open project...`.

0 commit comments

Comments
 (0)