Speech synthesis on Ubuntu

Text-to-speech (TTS) has been around for a couple of decades, and it keeps getting better. There are a bunch of really fun untapped applications for it, combining RSS, filters (like Pipes), podcasts, telephony, and hidden speakers.

Under Linux there is a nice package available called Festival. To get started, grab an appropriate package, such as:

• festvox-hi-nsk (Hindi male)
• festvox-kallpc16k (american English male)
• festvox-rablpc16k (British English male)
• festvox-mr-nsk (Marathi male)
• festvox-suopuhe-lj Finnish female
• festvox-suopuhe-mv (Finnish male)
• festvox-te-nsk (Telugu male)

Too bad you can't get a female speaker except in Finnish. (I had never heard of the Indian languages Marathi and Telugu, and I consider myself a language buff... sigh.)

The results are pretty good. Here's how to use it from the command line:

text2wave text-file.txt -o audio-file.wav

For extra fun, use pidgin-festival to turn incoming instant messages into speech (use festival-gaim if you haven't made the jump to Hardy Heron yet).