Skip to content

Imaginary Machines: eDJ

December 13, 2010

Aw who am I fooling; cutting back on the computer poetry doesn’t help my productivity any; I just end up playing more kdice and freeciv.

Anyways, this is another “what I might do if I had all the time in the world” system. I was watching some guy talk about his animated poetry, and I thought: what would I do if I were obliged to give a presentation about computer poetry? So basically I took parts of ePoGeeS and re-arranged them such that they could be used to generate text-to-speech-delivered (TTS-delivered) poetry in real time. Language modeling and generation are done at the far left of the interface. After lines are generated, they can be sent to an output window where they can be spoken automatically, or further edited after a length-of-utterance calculation is made. This is done so it can be coordinated with sound loops that will be playing in the background; you can fade between sound loops, or vary the TTS and sound loop volumes.

I probably won’t ever implement eDJ, but it’s another “fun to think through” exercise.


p.s. DJs are teh leet.

Advertisements
2 Comments leave one →
  1. eRoGK7 permalink*
    December 14, 2010 6:43 pm

    I want an “eDJ” program. I had thought of a performance of a poem that used live DJing of input text in conjunction with some kind of noise/dubstep music.

    We can always dream…

    • eddeaddad permalink*
      December 14, 2010 7:35 pm

      > I had thought of a performance of a poem that
      > used live DJing of input text in conjunction with
      > some kind of noise/dubstep music.

      Text-to-speech prosody is the tough part. You can manipulate TTS to make it sound better with punctuation and SSML, but it only helps so much. In the eDJ interface design I made two separate textareas, one for generation and one for TTS editing so that you could play the generated TTS to a sound out that only the DJ can hear before sending out to the audience.

      Anyway, if you want to try out real-time poetry generation for performance, you can just combine a couple existing apps.

      • google “cepstral tts”, download their free SwiftTalker app and buy a voice for $40
      • play a beat in Audacity or some other sound app
      • generate poetry in realtime using Gnoetry or some other generator
      • as you generate poetry, copy-and-paste it into SwiftTalker so it will be spoken over the beat

      That’s pretty much what I was doing with the youtube-delivered TTS poems on this site, I just wasn’t doing it in realtime.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: