speakgoodchinese

[newyorkeric]

I found this cool software at http://www.speakgoodchinese.org/index.html. It will record your pronunciation and test how good your tones are. Has anyone used it or is anyone willing to try it and supply some feedback on how well it works (it's free)?

EDIT: I should add that I tried it and liked it but am not sure how good the program is since I am a real newbie when it comes to Chinese.

[newyorkeric]

I asked a native speaker to try this software. IMHO, given the analysis of her tones, I think this software can be a great tool for beginners. I learned a lot about what I do wrong, not only regarding the tone shape itself but also about how I don't keep my pitch smooth, that I have trouble with two first tones in a row (dropping the second first tone below the first), etc.

I'm still curious about what others think.

[cdn_in_bj]

Looks neat, I'll give it a try this weekend and let you know.

[asialinks]

interesting... i tried it out too!

[roddy]

Any tips on getting it working? I've installed, but the two programs it adds to my Programs menu just open DOS boxes (is that the right word?) and then crash or close. There's a lot of info on compiling stuff on the website, but I was kind of hoping that wasn't intended for us humble Windows users . . .

[heifeng]

ditto here...lost on what to do after i get hit with that black box and an error message

[newyorkeric]

I'm not sure why it's not working for you. I am using a PC running Windows and there was no problem. It does open a DOS window first and then a second or two later opens a second window, which is the actual program.

For those of you who get it to work, there is an option for male, female, child speaker that you have to choose from the Voice pulldown menu so that the program takes your pitch range into account.

EDIT:

This is the version that I installed: http://www.speakgoodchinese.org/releases/sgc-install.exe which is listed under Windows download and installation instructions as Windows (right) icon on the download page.

[Skinkie]

Hello all

I'm the lead developer of the tool. Responsible for the Praat integration and crossplatform compatibility of SGC.

Currently we can say all developers responsible for coding (sgc/praat):

- Don't use Windows

- We are ready to help

While making this program we experienced some very strange problems in relation to praat and Windows XP. I have created two *totally* different implementations using praat (mentioned the compatibility update on our site). It became clear different machines responded *totally* different to the same code. If this was related to the praat program (which runs on virtually all platforms flawlessly... or is related to my powers to link C code...) was put in the middle.

...but we can help.

First I need to know the error message in the dos box, you can run sgc from the directory it is installed to. SGC *won't* work if you start it via the .exe(!) if you are not in the same directory. This is because it uses a .ini and .glade file. This could technically be fixed in very hard C code... so we decided we fix it using the wrapper file. Image program 'The Gimp' using this Win32 'where-am-I-installed-fix', so it could be implemented.

So I suggest for the people having problems:

Start > Run > cmd

c:

cdsgc

sgc

(Report the error messages)

One of your admins reported me this thread, you can find my email addy on http://www.speakgoodchinese.org/

ps. I don't know if anyone already published a new version on the site. We are virtually all on holiday after the official closing of the project (was hard work). But check out: http://www.speakgoodchinese.org/beta/

[roddy]

Yep, that was me - excellent response time!

Here's what happens as best as I can explain this. Please bear in mind that even starting up the command box is unusual for me

C:>cd program files

C:Program Files>cd SpeakGoodChinese

C:Program FilesSpeakGoodChinese>sgc

*********At this point I get the standard windows 'sgc.exe has encountered a problem' message, and nothing happens

then

C:Program FilesSpeakGoodChinese>singleword

Assertion failed: scaled_font->ref_count > 0, file cairo-scaled-font.c, line 607

This application has requested the Runtime to terminate it in an unusual way.

Please contact the application's support team for more information.

C:Program FilesSpeakGoodChinese>

Normal English XP Pro install.

[Skinkie]

Thanks, that doesn't look good. Because it is cairo (the basic layer - with GTK).

http://www.speakgoodchinese.org/beta/sgc-install-2007-07-27.exe

Can you check if this works, otherwise the only solution would be to redo the cairo compilation...

[roddy]

Same again, I'm afraid. Just about to try it on a different machine in case that makes any difference - it seems to be working fine for some lucky people.

[roddy]

Successfully installed on a different machine. No idea what the relevant differences are - both dell machines, a laptop (D420, didn't work) and a desktop of some variety.

If you can think what the problem might be and I can do any testing for you let me know - I'd like to see this working.

[Skinkie]

Some Microsoft fanboys should find out what is the difference between versions. It should be trivial to test which versions are broken in a real test bed, but WHY is it so broken :

[Skinkie]

Clues I have:

One of of the assistant-professors suggested 2nd edition XP versions vs 1st. I found some strange issues with sounds cards.

There must be a way to compare dll versions between to running systems. If you know someone who is in to this...

[roddy]

Integrated sound on both, working one has XP Home Service Pack 2, non-working one is XP Pro, SP2. If anyone can tell me what to look at I'll look at it . . . . tomorrow . . . .

[newyorkeric]

She thought that the software did a good job of tracking her tones. What helped me a lot was seeing what her tones looked like compared to mine. Hers were smooth and steady and mine were wobbly and jumpy.

[Skinkie]

We had 3 native speakers in the project group.

Here: http://www.speakgoodchinese.org/articles.html the test results can be found between correct rates of native speakers.

The tool is primary intended to get a 'feeling' for Pinyin and pronunciation of a tone language. Since this is the most common problem in the education business, people can train these basics at home now.

Although it is interesting how smooth the tones are of native speakers, don't forget that you are not, and your *ears* are not trained to help the articular system. Other research at the University of Amsterdam for second language acquisition (L2 learning) has a hypothesis that listening regularly to the foreign language will help you to make distinctions and therefore be able to pronounce them. Thus: watch chinese TV/radio and use this tool to test if you understand the tone distinctions and are able to reproduce them.

ps. Newyorkeric: bumpy can also mean that there is a lot background noise... really this software (and probably no other) can work on a pitch level with noise in it.

[newyorkeric]

Skinkie,

I have a few questions:

Sometimes I have sharp spikes at the beginning and end of my recording. Do you think it is because I am starting and stopping my speech incorrectly or could it be due to the sensitivity of my microphone and the software? I wonder if the mic is picking up utterances and breathing that native speakers produce as well but that aren't displayed in the reference pitch.

Does the program test absolute levels as well as relative levels? That is, does it test that my 2nd tone rises high enough or that my 4th tone falls low enough? Or is it testing that my 4th tone falls enough relative to the starting point?

I read somewhere in the documentation that the program is generous with regards to giving correct responses so as not to discourage new learners. In future versions it would be helpful if the user could choose the tolerance level.

I think I found a bug. After recording, the program sometimes won't respond until I click the mouse outside the program window.

Thanks for your responses!

[Skinkie]

offtopic: is there a quote button somewhere?!

Sometimes I have sharp spikes at the beginning and end of my recording. Do you think it is because I am starting and stopping my speech incorrectly or could it be due to the sensitivity of my microphone and the software? I wonder if the mic is picking up utterances and breathing that native speakers produce as well but that aren't displayed in the reference pitch.

Using the 'play' button you can listen back your recording (as rescaled to 0db). If the rest of the curve looks the same I wouldn't worry It is probably the detection of no speech vs speech.

Does the program test absolute levels as well as relative levels? That is, does it test that my 2nd tone rises high enough or that my 4th tone falls low enough? Or is it testing that my 4th tone falls enough relative to the starting point?

As you might have noticed you can choose the type of speaker. While developing the developer of the detection functions (my teacher/colleague) pointed out that the differences were so small that personalized voice-pitch detection was not necessary. So *YES* it is relative to the choice of speaker. But it isn't significant enough to create errors

The rising and falling works with semi-tones and is based on the 'relative' value mentioned above... I can point out... that the detection is probably the only one out there that is really based on documentation how it *should* be. Although we are still open to suggestion from peers. Our program will be featuring on the InterSpeech congress.

I read somewhere in the documentation that the program is generous with regards to giving correct responses so as not to discourage new learners. In future versions it would be helpful if the user could choose the tolerance level.

This is true. The problem was the time constraint this program is build within. We had only a half year for it... instead of a year. Since the program is licensed under the GPL we hope that more people would like to help to enhance it. There are already people very interested to use the framework for other tone languages.

I think I found a bug. After recording, the program sometimes won't respond until I click the mouse outside the program window.

GTK+ bug... the most annoying bug is you should move the mouse cursor before you can click on a button that was disabled while recording. I'm really hoping this bug gets fixed soon ;)

Thanks for your responses!

You're welcome And we are very open to comments, after August we will set up a bugtracker.

[newyorkeric]

The single tone tests seem to work fine and are very insightful. The two tone tests either work poorly and/or my tone combinations are really bad and/or I am doing something wrong. In most cases I get red-line gobbly gook that I can't make heads or tails of. Any advice?

Are the reference lines also meant to indicate the length of the tone? If the red line is longer that the reference line does this indicate that I am holding my tones too long?