This site is a web-based text to speech synthesiser. You can paste up to 300 characters-- about one paragraph. Not bad quality. I could listen to a 1 minute passage 2 or 3 times comfortably for an idea of pronunciation. Works for several varieties of English, French, German, and Spanish
http://research.att.com/~ttsweb/tts/demo.php
I would like to find a similarly easy to use resource for other languages.
How and when do you use this Ed?
These sites are similar in terms of languages and quality of voice. They force you to sign up and are a little harder to interface with. However they allow you to record much longer passages-- up to 5242880 characters and there is a Firefox addon that allows you to select text, right click and create a surprisingly listenable recording.
http://www.spokentext.net/s_index.php?lang=en
PS a LingQ Firefox Add-on would make my day, or even my week!!
Currently I don’t. I just thought it would interesting to put out there. Particularly those learning English since it is such difficult language to pronounce based on text. I think learners of Chinese would be in the same boat although I am not sure. I see it as an aid to those who like to have a quick idea of how a passage that they randomly come across sounds.
I do not like listening to synthetic speech as a learning tool but am curious to hear the views of others. I,for example, would rather pay for a well narrated audio book than listen to a poorly narrated free audio book. Dictionaries offer the pronunciation of individual words. But to get rhythm of a language, I find it is best to hear lively natural language. In fact I find that the resonance of content is important to our ability to pick up on the feeling and mood of a language. As I have said before, in my Chinese studies I found certain tapes, like certain comic dialogues, particularly good for picking up tones.
I often think that if we could get content from Vinyl Cafe, that would be great for picking up on the rhythm and flow of Canadian English for example.
But there may uses for synthetic voice that I am not aware of.
At any rate it is always interesting to see the many new language related systems and resources that are out there. Thanks for this and thanks for fixing up the spread sheet with sources of content, which is growing.
Maybe more useful for LingQ would speech-to-text. Does such a thing exist? Tech people?
I think, Edward, that the short answer is, yes but it’s rubbish! I’ll let a tech person give a more detailed answer…
I like http://imtranslator.com/ and it has a text to speech function for most languages including Russian.
I’d rather not listen to machine generated speech if I had the option of listening to real people though. I’m with Steve on the good audio books soapbox.
Fantastic! I especially love the no nonsense interface.
Well, I find this synthesizer interesting and a bit useful for learning. Thanks Ed.
I don’t like too much synthesized text either, mainly if we are speaking about leaning a language. But if there is something out there that you want to study but can’t find a human recording for it, I think it is better than no audio at all.
Speech to text is a hot research topic in Artificial Intelligence, but it is yet rather difficult to do it with a decent error rate. Most systems are able to identify only isolated words or short expressions. This can be useful, for example, to build voice-based interfaces, in which the system would recognize simple predefined commands.
Here is an example http://www.voicesearchbar.com/ of what I mean.
Google is promising for soon a service of voice mail transcription, as you can see at Voice.
I’m really looking forward to see how this will work!
I haven’t used TTS much, only a couple of times when I’ve had short texts without audio. TTS may not be the real thing, but probably better than my own attempts at sounding out the words - as Ed says, “for an idea of pronunciation”.
Speech to text, if it were accurate, would be awesome and would open up a whole world of podcasts. However, it hasn’t proven to be accurate up till now. With Google looking into it, that sounds promising!