ISSE User Manual: Version 0.2.0 (Alpha) The ISSE Team October 14, 2013
ISSE User Manual: Version 0.2.0 (Alpha) The ISSE Team October 14, 2013
ISSE User Manual: Version 0.2.0 (Alpha) The ISSE Team October 14, 2013
isse.sourceforge.net
1
1 About
ISSE is an interactive sound source separation editor. It allows you to import a single
audio recording and separate it into two distinct sound sources via drawing and painting
tools. Example tasks include separating a cell phone ring from recorded speech, separating
drums mixed with bass guitar, or separating vocals from background music. See our demo
video for more information.
The software is an open source, freely available, cross platform audio editing tool devel-
oped by the Center for Computer Research in Music and Acoustics (CCRMA), Stanford
University and Adobe Research. It is licensed under the GNU General Public License,
Version 3. Authors include: Nicholas J. Bryan, Gautham J. Mysore, and Ge Wang.
For questions or comments, please see our
• Main web site http://isse.sourceforge.net/
• User forum http://isse.sourceforge.net/forum/
• Wiki http://isse.sourceforge.net/wiki
• Mailing list http://isse.sourceforge.net/contact.html.
2 Installation
2.1 OSX
2.2 Windows
2
4. Note, additional materials (test files, manual, etc.) are also installed along with the
application.
5. Done!
2.3 Linux
3 Quick Start
1. For a quick introduction to the application, please see the demo video posted at
http://isse.sourceforge.net/index.html.
2. Open the application and start a new project via: File → New.
3. Select a short (5-30 second) .wav file. Longer files are permitted, but will be slow to
process. If needed, select one of the example files from the Test Files folder of the
download package.
4. View and listen to the sound. Time goes from left to right. Frequency goes from
bottom to top. You can zoom, scroll, etc. Click the play button at the top or hit
spacebar to listen.
5. Choose which sound you would like to separate (e.g. vocals vs. background music,
cell phone ring vs. speech, etc.).
6. Associate one of the two colors from the top left to a given sound (e.g. vocals). Asso-
ciate the remaining color for the remaining sound in the recording (e.g. background
music).
7. Start annotating the visualization of sound. Select one of the drawing/painting tools
(e.g. brush, time select, frequency select, box select). Draw on salient regions of the
visualization that correspond to each sound source (e.g. drums, bass, vocals, etc).
3
Use the brush controls (click the plus sign to right of the brush tools) as a measure
of confidence. You do not have to annotate the entire file.
8. Click the process button (upper right). Wait a little bit and the middle and bottom
audio tracks will soon display the separated results. When the processing button is
on the results will continually update in the background.
9. Listen and view the separated results. They won’t be perfect, but...!
10. Repeat the annotation process. Now draw on the input and output to refine the
separation.
11. Export the separated results via: File → Export Audio and save the project via:
File → Save As. Note, to save, export audio, open, and create new projects, the
processing button must be off and the processing progress bar must 100%.
12. Done!
4
ISSE - An Interactive Source Separation Editor
Drawing Control
Display Control
Extra Brush Controls Process Master Volume
Source Select
Loop Markers
Source 1 Annotation
(Blue)
Separated Source 1
(Drums)
Source 2 Annotation
(Red)
Separated Source 2
Bass
Layer Visibility Controls
Playback Cursor
Auto Solo Mode
5
4 Screenshots
To give a better idea of the functionality of ISSE, please see the screenshots below.
Figure 2: The Multi Paint View. In this view, you can see both the input audio file and
the two separated output sounds.
5 Technical Documentation
Technical publication on the source separation algorithm use within ISSE can be found
at:
• [?] - An extended abstract and introduction to the idea.
• [?] - The core algorithm and technical material.
• [?] - Algorithmic extensions and additional interpretation.
6
Figure 3: The Single Paint View. In this view, you get a larger screen space to view a
single track. Moving back and forth between the Multi Paint View and the Single Paint
View can be done quickly by right-clicking on a track and selecting the popup menu.
7
Figure 4: The Separation Settings View. In this view, you change parameters of the sep-
aration algorithm. Careful, changing certain settings can clear your painting annotations
(an alert box will tell you which ones...).
8
Figure 5: The Visualization Settings View. In this view, you change parameters that
control the visualization of sound such as the color map and the clip limit.
9
Figure 6: The Brush Settings View. In this view, you change additional painting brush
parameters. Click the plus sign to the right of the brush icons.
10
Figure 7: The Audio Settings View. In this view, you can change the audio settings (sample
rate, buffer size, etc.). To open this view, go to File → Audio Settings.
11