GUI text-based speech and music editor for creating radio/audio stories



Enabling the music browser

Issues and bugs

Basic interface usage

Step-by-step editing tutorial

Adding speech tracks tutorial

Podcast producers


View the Project on GitHub ucbvislab/speecheditor

Edit speech like it's text

The speecheditor is a web app for editing and adding musical scores to audio stories. Instead of editing speech using waveforms, the speecheditor allows you to edit audio with text, like a word processor. Likewise, it allows users to easily loop musical tracks and add musical underlays à la This American Life.

The speecheditor was developed in the VisLab at UC Berkeley in collaboration with Adobe Research. You can find the paper here: http://vis.berkeley.edu/papers/audiostories.


Download and install Vagrant

Download and install VirtualBox (or possibly another Vagrant host like VMWare Fusion).

Clone the speech editor repository:

# (> means a prompt on your computer)
> git clone https://github.com/ucbvislab/speecheditor.git
> cd speecheditor
> git submodule init
> git submodule update

In the /speecheditor directory, spin up the vagrant enivornment by running

> vagrant up

This will take time: vagrant will set up a virtualmachine with all the requirements necessary to run the speech editor.

Now, to run the speech editor:

# ssh into your vagrant box
> vagrant ssh

# the /speecheditor is mounted on the vagrant box at /vagrant, so cd to there
# (the $ means a prompt inside the vagrant box)
$ cd /vagrant

# run the speech editor!
$ python app.py

To access the speech editor, go to http://localhost:8080 in Chrome only.

See the tutorial for how to use the speech editor at http://localhost:8080/static/tutorial/index.html.

You are free to edit the source code on your computer; everything will be shared between your main computer and the virtual vagrant box.

If you change the javascript or coffeescript, run

$ grunt dev

inside the vagrant box to regenerate the javascript.

You can exit the ssh session with

$ exit

Once you're done using the speech editor, free up the system resources taken up by the vagrant box by running:

# from the /speecheditor directory on your computer
> vagrant halt

Then, vagrant up, vagrant ssh, cd /vagrant and python app.py to start it up again.

Enabling the music browser

If you have access to the music browser data files (private access only due to copyrights): put the musicbrowser folder inside of /speecheditor/static. Also put the music_browser_app folder inside of /speecheditor. Instead of using python app.py run the speech editor with:

$ python app.py --music-browser

Issues and bugs

Please file all issues and bugs in github issues.

Got speech tracks?

If you are a producer/editor/podcaster and you have speech tracks that you would be willing to share with us, please let us know! Short tracks are fine. Email maneesh at cs.berkeley.edu and srubin at cs.berkeley.edu. We ask that the speech be public domain or a permissive Creative Commons license. In return, we'll get your speech track set up to be edited in our system.