Rudolf Kastl: >>> personally i found that http://espeak.sourceforge.net/ is already >>> much more versatile. Tim: >> Hmm, neither this, nor Festival, sound as good as the speak device on >> the Amiga (from the mid 1980s). That's a bit surprising, I'd have >> thought that it'd be vastly outperformed, by now. Ed Greshko: > Is this observation from memory or do you have an Amiga running today and > are making an actual comparison? Just wondering. The Amiga's speech sounds quite like Steven Hawking's speech synth, not to mention a few other speech synths used by the disabled. That'll give a few more people an idea what it sounds like, seeing as it pops up in movies and television. I couldn't easily drag mine out and sample it, for quite some time, it's buried amongst the rest of my clutter. While it doesn't sound very polished, it is quite understandable, and handled most words and sentences fairly appropriately (changing pitch, and pausing, where it should, etc.). There were some words that it spoke badly, but that's not unique to speech synthesiers, and it did have a way of you giving it a library of words to be pronounced according to your own rules (applied to the synth, systematically, rather than needing to customise whatever program you were playing with). The English language is full of words that break the rules, so I'm not surprised at this, and often wondered with speech synthesising in other languages is more effective. I found the Festival one to sound rather clean (not grungy) but rather monotonic and very difficult to comprehend. I'm reminded, somewhat, of some rather old children's cartoon that had this bird which wore half an eggshell like a hat, that spoke every word on the same high pitched monotone. It's voice could set your teeth on edge. I listened to the provided ogg file on the eSpeak website, I found it very nasily, over modulated for each word, making me think of bubbling something up from underwater, and very little correct modulation of words throughout a sentence. Rather like listening to someone read out loud who was very poor at it, speaking all the words at the same rate, no pauses in the appropriate places, and no raising of pitch at the right moments (questions), etc. I have heard better ones than any of them, one being something on Windows 98 SE that made use of those Microsoft Agents (Peedy the parrot, etc.). I could use it to speak the text from IRC, and follow the conversation quite easily, even when I ramped up the speed a bit above normal speech. I do have fairly fond memories of playing with the Amiga's speech synthesier, it was quite easy to supply it various parameters and change how it spoke from word to word. So I did the obvious, made up a little program that made it sing a song. 'twas done using just BASIC, but still amusing. -- [tim@bigblack ~]$ uname -ipr 2.6.22.1-33.fc7 i686 i386 Using FC 4, 5, 6 & 7, plus CentOS 5. Today, it's FC7. Don't send private replies to my address, the mailbox is ignored. I read messages from the public lists.
