I have been looking for information that will help a high school student who is deaf 'listen to' her classroom teacher in her non-native English. In her native Japanese she does very well at lip reading, but is still developing the skill for use with English. While not perfect, I think I finally have something that might help her that is fairly low budget. If you all have low budget ideas to improve speech recognition in a mostly oral, teacher intensive learning environment for a student who is deaf and learning in his/her second language, I would love to hear from you.
Hardware:
- Android 5.0 device that allows external microphone input. She uses a Sony Xperia Z2 Tablet. (already had)
- Bluetooth microphone. I got her the Sony ECM-AW4 (about $150)
- Adapter to allow input of the microphone into the headphone jack. In her case (M) TRRS 4 pole to a (F) 3 pole mic and (F) 3 pole headphone jack (which is not needed for only listening) (about $6)
- Cable (M) 3 pole to (M) 3 pole, to attach the receiver to the tablet. Included with AW4.
- Total expense was under $175 with no extra monthly... charges.
Software: Android app, Listnote. It is free in exchange for seeing adds. I wish they had a pro, non-ad version as well.
Usage:
- The instructor wears the clip-on microphone which sends his voice to the receiver. Power on.
- The student attaches the audio output from the receiver to the Android device. Power on. The bluetooth connection which apparently has a 100ft line-of-sight range connects almost immediately. No pairing... needed. It is super simple. Turning up the volume switch seems to increase accuracy.
- In settings, set the Continuous Speech setting ON, Quick note Timeout to MAX and Continuous Speech Timeout to INFINITY.
- In the main page Listnote click Speech Recognition. The page changes and a yellow bar times out in 10 seconds. If text is spoken, it should show in that box. Then if it times out you will be able to click the Continuous Speech box. When you do so, you no longer need to worry about timing out. It just continues to listen and enter text.
Note: In that Continuous text mode, if there is a little pause, the speech detection tends to miss a word or two, then liks in again. I am guessing that watching for breaks and clicking Enter to create a new line will assist in later going back and reading the notes. It does not detect the beginning or end of a sentence. It is not perfect, but it is for sure better than nothing and the best I have found.
Another possible feature. The ECM-AW4 also allows speech to flow from the student to the teacher. So the student can enter text in an app such as Google translate and press the speaker to verbally respond to the teacher, if the teacher and student desire to do so. Her verbal English is currently about 70-80% recognizable, so I am not expecting that portion of the technology to be use.
