backtop

Print E-mail del.icio.us 54 comment(s) - last by Shoal07.. on Jan 3 at 2:49 PM
Reading and listening to DailyTech just got a little easier

It won't be too often that I post shameless self-promoting articles on DailyTech as top-tier stories, but if it was ever going to happen today's the day.

When DailyTech originally launched, it was always part of the site plan to include some form of audio-enabled content.  At first we attempted to use human readers. This proved too costly and too slow to create content accurately.

In 2006 we attempted the project a second time with AT&T's Text-to-Speech API. However, even at a mere 10 million page views per month, a moderate text-to-speech license ran in the five-figure range.  Not exorbitant, but certainly a hefty investment for something our readers might not even like.

Enter Odiogo.  The premise is simple: Odiogo takes an extended news feed from the content provider, runs the content against its text-to-speech software, and then republishes it as an XML feed.  This XML feed can be integrated into an RSS feed for a podcast, or directly into the article as a "play" button. 

Nobody will admit Odiogo is a replacement for human-read news -- yet.  The software is far from perfect when it comes to subtle pauses, tech jargon and the odd acronym.  But considering the difficult nuances of the English language, and the improvement of text-to-speech software in just the last two years, I'm pretty stoked about Odiogo and text-to-speech in the future.

Give DailyTech's Odiogo feed a try.  You can play audio directly from the website on the "listen now" button, or you can listen to rollups of the news via podcasts for iTunes and other MP3 players.  If you like (or don't like) what you hear, drop us a comment!

Comments     Threshold


This article is over a month old, voting and posting comments is disabled
Interesting
By Gnoad on 12/29/2007 12:32:53 PM , Rating: 3
It's amusing, and potentially useful, but I doubt I'll end up using this much. I tried listening to a few articles, and after that the voice was starting to irritate me. I think this kind of software is always getting better, but it still has a long way to go before it can replace humans.




RE: Interesting
By James Holden on 12/29/2007 12:34:06 PM , Rating: 5
I wonder if they have a sexy female voice ... that seems like it would do better in this crowd :) Maybe with a foreign accent too!


RE: Interesting
By tdawg on 12/29/2007 1:31:43 PM , Rating: 2
There used to be a morning show on a radio station here in Seattle that did this sort of thing with the news one morning. The female half of the show took local news and science articles and the like and read them in her best sexy, sultry voice, which was absolutely hilarious.


RE: Interesting
By KristopherKubicki (blog) on 12/29/2007 1:38:20 PM , Rating: 2
Chicago's local cab company (Flash Cab) uses TTS for its reservation service. For some reason the female TTS voice sounds like it was ripped from a phone sex hotline or something ... I always get a chuckle out of it.


RE: Interesting
By tdawg on 12/29/2007 2:34:58 PM , Rating: 4
That's the voice she was doing the news in. I almost crashed my car listening to it. :)


RE: Interesting
By shadowofthesun on 12/31/2007 4:33:31 PM , Rating: 2
Both hands on the wheel, tdawg. ;)


RE: Interesting
By therealnickdanger on 12/31/2007 1:19:10 PM , Rating: 2
Sometimes just for fun, I change the voice in my car from English to French... she sounds so hot. Reminds me of the SNL skit involving the "Buick Mistress". I'm not sure if that was the exact title, but it was such a great skit.


RE: Interesting
By Spyvie on 12/31/2007 5:48:43 PM , Rating: 2
You could always download the trail version of Read Please, then highlight any text you want and have it read aloud in one of several male and female voices. I've been using it here and there for years, but it doesn't sound quite as good as Odiogo is sounding here...

http://www.readplease.com/english/downloads/

I've never tried MS Sam, it may be the same thing.


RE: Interesting
By MrDiSante on 12/29/2007 1:51:19 PM , Rating: 3
I think this is a fairly impressive achievement. My dad couldn't tell that it was a robot for the first paragraph or so. Compared with something like Microsoft Sam, this is a huge improvement.


RE: Interesting
By Cogman on 12/29/2007 7:44:46 PM , Rating: 2
Maybe its just me, but I thought it sounded exactly like Microsoft Sam. It is definitely a step back from AT&T's tts API.

On the other hand, I think it is a good addition to daily tech, and should benefit the blind and the lazy users :) Good Job those that made this work.


RE: Interesting
By Cogman on 12/29/2007 7:46:43 PM , Rating: 2
I take that back, only at low volumes (with load background noise) does is the computerized voice obvious. When I turned up the volume it was strangely less obvious (and sounded less like my beloved Sam)


RE: Interesting
By jtesoro on 12/30/2007 1:29:30 AM , Rating: 2
Now I wonder what will happen if this capability were applied to the comments and not just the articles. I don't even know what its supposed to say when it gets to LOL, w00t or STFU! :) <-- Yeah, and that smiley too!


RE: Interesting
By KristopherKubicki (blog) on 12/30/2007 10:58:24 AM , Rating: 3
This technology will never be used for our comments. We do want some people to still read the site ;)


Underwhelming?
By ice456789 on 12/29/2007 12:12:06 PM , Rating: 3
So I clicked on the button to listen to the article and I got a message that said "Service not available yet". But I didn't pay anything for it so I guess I got my money's worth.




RE: Underwhelming?
By KristopherKubicki (blog) on 12/29/2007 12:18:33 PM , Rating: 2
It take's a couple minutes after an article is posted. It's working now.


RE: Underwhelming?
By TomZ on 12/29/2007 12:52:13 PM , Rating: 2
The delay might be something of an issue - seems like the other two new articles don't have audio available yet. Any idea what the actual delay time is? I suppose it's more than a couple of minutes. One of the articles was posted at 11:01am (CST?) which was about 50 minutes ago, still no audio. I don't know how accuate those timestamps are, though.


RE: Underwhelming?
By KristopherKubicki (blog) on 12/29/2007 1:09:01 PM , Rating: 2
It's supposed to be about a 10 minute delay, though we have the ability to produce the TSS on-demand from our content management system -- that hasn't been deployed yet though.

And yes, the timestamps on our articles are somewhat whimsical. It's a bad habit.


RE: Underwhelming?
By ice456789 on 12/29/2007 2:55:20 PM , Rating: 2
Now that it's working, I can say that I really am amazed. I was expecting the typical burping robot voice.

Johnny 5 is alive!


RE: Underwhelming?
By Some1ne on 12/29/2007 6:54:26 PM , Rating: 3
Works fine for me. It's not perfect, mostly because it lacks various subtle nuances that are part of real human speech (like for example, placing emphasis correctly on conceptually relevent terms within a sentence), but for a completely automated, synthetic voice, it's pretty damn impressive.


By BruceLeet on 12/29/2007 12:30:44 PM , Rating: 2
I can read/process information much faster, maybe add a speed button? Because at that speed I forgot what he said at the beginning of the sentence nearing the end of the sentence and it just doesn't work for me, plus news articles need to be "proof-read" especially on a site like this where much debating occurs, don't really see it being anything all that great, but thats just my onion :D

Ye I know I spelled opinion wrong, take it easy grammar nazi's




By JackBeQuick on 12/29/2007 12:32:56 PM , Rating: 2
Yeah I think if you can read it, reading is faster and easier. This seems more geared to people who take their news on the go or want to listen to the site in the background.


By johnsonx on 12/29/2007 3:42:58 PM , Rating: 2
quote:
Because at that speed I forgot what he said at the beginning of the sentence nearing the end of the sentence


A.D.D.?