Wednesday, November 7, 2012

How to use free text-to-speech software for dialogue completion tasks

YAKiToMe

Do you use text-to-speech software with your students? 

  • There are some really good software packages these days, such as Nuance's Dragon Naturally Speaking--however, the services with high quality voices usually have a price tag.  
  • http://www.yakitome.com/ one free site that I use that has reasonable quality voices.  It also has options for other European languages, like Spanish, German and French.

This free text-to-speech engine runs in the browser.  
  • You can also download an MP3 created from the text you type.  
  • In English, you can use UK or US accents, and select female or male voices.  You can also have the speech created at normal or slow pace.  
  • There are some simple 'voice tags' you can use within the transcript to change voices (if you want to create a dialogue), add pauses, or adjust the speed.
Why use TTS?

As part of the curriculum, students here at METU NCC are expected to complete gaps in a dialogue.  Students can get a bit bored of doing this on paper, so I thought that YAKiToMe would offer a nice alternative.

How to use TTS in class?

One idea I had would be for the students to work in groups and create 'one side' of a dialogue as text.  
  1. Then, they use YAKiToMe to create the 'one sided' dialogue as an MP3 audio file.
  2. Groups then exchange the MP3 dialogues with another group. 
  3. In a computer lab, the group then uses AUDACITY to edit the 'one sided' dialogue, and insert their responses for the 'other side' of the dialogue.  
Alternatively, if you only have one computer in the class, once a group has created a 'one sided' dialogue, you just play the MP3 in class and stop it after every line.  
  1. Students then transcribe the audio individually and then compare their transcription with a partner.  
  2. Together, the write the complete dialogue.  
  3. Afterwards, they can do a role play and you can discuss the variety of responses given.

Sustainability?

I think that this would be an interesting way to practice writing dialogues.  But, what is more, once your students are aware of how to create audio from text, you could cajole them into creating their own 'talking books' class library, and the students could then use the site on their own to create and share 'talking books' for listening practice.

Just as an example, here is the dialogue I wrote:
You will hear 'one side' of a conversation with Lauren. Edit the MP3 file and use AUDACITY to complete the conversation with you as the person Lauren is speaking with. Afterwards, delete these instructions and set the scene for the conversation for the listener - who are the characters, where is it taking place, etc.
Hello.
My name is Lauren. What is your name?
I'm fine.
My favourite colour is black. I am a bit of a sadist. What about you?
That's interesting. Why?
I'd love to, but I have to run.
Yes, that's a good idea. Give me a ring.
Bye!
I then converted this text to audio using YAKiToMe.  I published this as a podcast below.

If you want to try out the procedure, download the MP3 and edit it on AUDACITY.  To do this, go to the EPISODE in PODMATIC by clicking on theYAKiToME podcast title.  At the bottom of the episode podcast you will see a DOWNLOAD EPISODE option. Once downloaded, you can open it in AUDACITY for editing.
 Upload your final version as a podcast and share it here or on TWITTER.



The use of computer-generated language tools is a relatively new development.  No doubt there will be more developments on the horizon.  Consider the text-to-speech (TTS) application and reflect on the issues and challenges this may present to teachers and students.  Here are a few issues that come to my mind:

  1. Would I recommend this to students for listening or speaking practice?  Why/Why not?
  2. What objections might teachers and students have about TTS?  How would I respond to such objections?
  3. Will the evolution of computer generated language change the why and wherefore of teaching English as a foreign language? 
If you are keeping a blog for reflective practice, create a blog post with your musings.  Share it with your PLN.

2 comments:

  1. I found a way to resolve this by making text-to-speech which will be easier for readers to look for content.

    ReplyDelete
  2. The current best text to speech software is Text Speaker. It has customizable pronunciation, reads anything on your screen, and it even has talking reminders. It is great for learning languages as it highlights the words as they are being read. The bundled voices are well priced and sound very human. Voices are available in English, French, Italian, Spanish, German, and more. Easily converts blogs, email, e-books, and more to MP3 or for listening instantly.

    ReplyDelete