Japanese Text-to-Speech Engine

So I was sniffing around Google, looking for something interesting to read, and came across this article at a place called ネットマニア(NetMania). It would appear that Pentax are in the process of developing a very good, natural sounding text-to-speech engine. And here I thought all they did was make cameras — as it turns out, they may be planning to integrate it into digital cameras in some way, and of course they could license the technology. Anyway, you can test drive it here at the Pentax website. Just copy and paste some Japanese text into it and be amazed…Or, impressed; the engine does make occasional mistakes (like reading “先ず(まず)” as “さきず”), but those are forgivable, and it does sound pretty good. It might be of some use to you in practicing Japanese.


This blog post was brought to you by the generosity of AJATT's patrons: Luke, Charlie, Nathan H, Other Nathan, Kyle, Aujury, Riad, Robert, memo, Nico, RK, Phillip, Mike, Henry, William, DaiSaka, Russell, remy, Adam, Michael, Jinette, Josh, Kent, Elin, Mairo, Christian, npkdyrpubfr, an.selenium, Squishyface, Diogo, Jeffrey, Nicholas, Wong, Toucan, vvv, Stefano, Chris, TMeurs, David, Neito, Quinn, Roodolph, Roger, dm, Lukas, Nenjya, Tom, Daniel, Francois, Richard, Amir, Matt, Hadi, Jace, Jean-Felix, Luke, Stijn, Nicole, Walter, Ian, nathan, May, Nyagasaki, Daniel, Emily, Coolbgdog, Cush, Erin, Stian, Christopher, Celia, SoloTravelBlog, Rob J, Jan, Tony, Avtar, Angela, Allen, Analisa, Eric, W, emk, Radek, Zach, Matt, William, Sarah, Jamie, LS, Nico, niin, Russell, Tawfiq, Jenny, Caleb

You guys are the best and I want to have your babies.

So...Let me have your babies.
Please? :D

Wait, what? All uterine humour aside #StopUterusShamingMeBro, seriously, your support means more than you know and I (Khatz ← that's me) am deeply grateful for each and every one of you. Thank you so, so much. Thanks for believing in me, thanks for taking action, thanks for being there.

If you would like to support the continuing production of AJATT content as well as two adorable cats (yes, actual cats) please consider making a monthly donation through Patreon. Right there. Go on. Click on it.

  8 comments for “Japanese Text-to-Speech Engine

  1. Saru Sponge
    May 9, 2007 at 19:46

    Very interesting. The English is still a tad stilted, but it is certainly an improvement. How does the Japanese compare? To me, a beginner, it sounds fairly natural. How is it in reality?

  2. khatzumoto
    May 9, 2007 at 19:52

    Just like you said, an improvement. If you put in a single sentence in Japanese, like “テメエ、打っ殺すぞ”, then you can’t tell the machine from a human being. But when you put in a whole paragraph, it starts to sound odd in certain places–the pauses are a bit off, etc. Not perfect, but definitely a step up.

  3. Saru Sponge
    May 9, 2007 at 20:00

    Well, the future is now. Soon we’ll be driving around in hover-cars and shooting lasers from our fingertips. Ha!

  4. khatzumoto
    May 9, 2007 at 20:04

    I know! When I heard the machine talking, I had these visions of “Star Trek” and universal translators whizzing through my head…Time to go and read some more Ray Kurzweil.

  5. AJ
    March 19, 2008 at 05:11

    Hey Katz,

    Just thought you’d like to know that the link here doesn’t work anymore. I tried to find a different one, but I couldn’t. Do you know the status of this project?

  6. khatzumoto
    March 28, 2008 at 10:07

    Negative. But I’ll do a more detailed post on TTS at some point. I’ve been using it in my studies/play/whatever with great success.

  7. Albe
    March 28, 2008 at 23:14

    I am using your 10,000 sentences method with TTS and I think it’s very useful. I read and listen the sentences, then I repeat.

  8. omerta
    July 28, 2012 at 18:34

    I’m just wondering if there would be a way to make a bookmarklet similar to that of the nonstoptube and google bookmarklets in which i could highlight text, click the bookmarklet and have that highlighted text appear in the voicetext.jp/ text box with a Japanese voice already preselected. Then it would just be matter of mining a sentence, listening to it and moving on.
    if anyone knows how to go about making such a thing, please let me know… or you could just, ya know, make thus said little time saver and pass it on 🙂

Leave a Reply

Your email address will not be published. Required fields are marked *