News

Project Updates

09/05/2010 - MTVR beta version 1.0.0.38 has been released and is available from the website. This is a minor update to our recently released 1.0.0.37 upgrade that addresses two issues: a tendency to hang on audio recording with Windows 7 64 bit systems; and failures to calibrate when there was a small amount of low frequency hum due to electrical interference in the audio input.

If you have already upgraded to version 1.0.0.37, we recommend upgrading to this version. If you did not upgrade to 1.0.0.37, you should skip upgrading to this version as well.

08/26/2010 - MTVR beta version 1.0.0.37 has been released and is available from the website. If you are using an earlier version and have not yet begun recording the full inventory, we strongly recommend that you download this version and create a new inventory (you will need to redo the screening sentences). Please make sure you completely uninstall any older versions of MTVR first before installing 1.0.0.37.

If you are already in the midst of recording your full inventory with the previous version of MTVR and it has been working well for you, it is best for you to finish your recording without upgrading to this new version of MTVR. If you have any questions or concerns about whether to upgrade or not, please contact us at beta@modeltalker.com.

Among other changes described below, with the 1.0.0.37 release of MTVR, we are replacing the original recording inventory of 1649 utterances with a completely new set of 1600 sentences. Unlike the old inventory that contained some very unusual sentences, and quite a few nonsense words embedded in short phrases, the new inventory consists entirely of short meaningful sentences and phrases drawn from publicly available texts. These sentences should be much easier to record without uncertainty as to how some words ought to be pronounced.

MTVR 1.0.0.37 contains many bug fixes and improvements. The following is a brief summary:

1. Calibration - The new version of MTVR has more stringent requirements for both the total amount of background noise that is acceptable, and for the Signal to Noise Ratio, that is, the relative loudness of the recorded speech when compared to the loudness of the background sound level in the recording environment. We have also improved the consistency and accuracy of the sound level measures MTVR performs during calibration.
   
   
2. Usability - We have made many small changes to the wording of instructions, the text appearing in buttons, and overall consistency in the design of dialog boxes and controls to improve appearance and clarity of the program controls. These changes extend to the sequence and flow of events to make operation more transparent.
   
   
3. Stability - We have identified and fixed many bugs that could lead to MTVR crashing, hanging, or behaving erratically under some circumstances.
   
   
4. Compatibility - As of 1.0.0.37, we have an experimental version available for use with 64-bit Windows 7 systems. Please send an email if you would like to try this version as it is not presently included in our regular installer.

02/18/2010 - MTVR beta version 1.0.0.36 has been released and is available from the website. The installer is now completely compatible with Windows 7 (32-bit version only).

01/05/2010 - The current versions of MTVR and ModelTalker are fully compatible with Windows 7; however, the current installer is not Windows 7 aware. A new installer and updated versions of MTVR and ModelTalker will soon be released to fix this problem.

08/14/2009 - MTVR beta version 1.0.0.33 has been released and is available from the website.

02/16/2009 - MTVR beta version 1.0.0.32 has been released and is available from the website.

11/14/2008 - MTVR beta version 1.0.0.30 has been released and is available from the website.

10/07/2008 - MTVR beta version 1.0.0.29 has been released and is available from the website.

08/15/2008 - InvTool has now been replaced by the ModelTalker Voice Recording Tool (MTVR). We believe this newly designed recording tool will be easier and faster to use and should be compatible with Windows Vista. It is still (early) beta software so the normal cautions apply. We will still support uploaded speech from previously downloaded versions of InvTool through the end of October 2008. After that time, please contact us directly if you still need to finish the recording process with the older InvTool.

02/08/2008 - Notification of fully funded second year of Phase II Technology Transfer (STTR) grant from the NIH National Institute on Deafness and Other Communication Disorders (NIH Grant Number R42 DC006193-04).

03/30/2007 - All new beta users of InvTool will be required to record a short 14 phrase inventory (included) and upload it before recording a full inventory. This is to insure that everything is working properly and the inventory is being correctly recorded. Once the screening inventory is checked and verified by our staff, we will send users the full inventory for recording.

01/01/2007 - Development of the ModelTalker and InvTool programs is now being supported by a Phase II Technology Transfer (STTR) grant from the NIH National Institute on Deafness and Other Communication Disorders (NIH Grant Number R42 DC006193-03). This grant will help us finish commercializing this technology from the Speech Research Laboratory of the A.I. duPont Hospital for Children and Nemours Biomedical Research.

Recent and Upcoming Presentations

2008

Bunnell, H. T., & Lilley, J. (2008). Schwa variants in American English. Paper presented at InterSpeech 2008, Brisbane, Australia.

Yarrington, D., Pennington, C., Bunnell, T., Gray, J., Lilley, J., Nagao, K., & Polikoff, J. B. (2008). ModelTalker Voice Recorder (MTVR) - A System for Capturing Individual Voices for Synthetic Speech. Paper presented at the ISAAC 13th Biennial Conference, Montreal, Canada.

Yarrington, D., Gray, J., Pennington, C., Bunnell, H. T., Cornaglia, A., Lilley, J., Nagao, K., & Polikoff, J. B. (2008). ModelTalker Voice Recorder – An Interface System for Recorded a Corpus of Speech for Synthesis. Paper presented at the 46th Annual meeting of the Association for Computational Linguistics: Human Language Technologies (ACL-08:HLT), Columbus, OH.

2007

Bunnell, H.T., & Lilley, J. (2007). Analysis Methods for Assessing TTS Intelligibility. Paper presented at the 6th ISCA Workshop on Speech Synthesis, Bonn, Germany.

Pennington, C. (2007). ModelTalker: Automatically Creating Personal Synthetic Voices. Presentation given at Messiah College Senior Computer Science Seminar, Grantham, PA.