Release Notes - 2018A

URL: https://github.com/nihospr01/OpenSpeechPlatform-2020B/tree/2018a_v1.0

This release introduces Speech Enhancement (SE).

SE is an advanced Hearing Aid feature for improving intelligibility in noisy environments. The baseline SE consists of:

Peak and Valley detection
Noise estimation
Spectral Subtraction using Wiener Filtering

Three approaches for noise estimation are provided:

Arslan Averaging Procedure
Hirsch & Ehrlicher weight averaging procedure
Cohen & Berdugo MCRA Procedure

The baseline SE is provided in source code.

Release 2018b introduces the ResearcherApp.

This app is used for controlling amplification parameters in real time using any browser enabled device. A video demo of the app can be seen here.

Additional web apps (e.g. NAL-NL2_app for generating initial prescription and configuring OSP accordingly; 4 alternate forced choice (4AFCapp), etc. will be coming soon.

We hope researchers will use these apps as templates, create additional web apps and share them with the research community.

We have started a Forum.

This Forum is for OSP users to discuss feature requests, bug fixes and knowledge dissemination. We thank you for your contributions to the Forum, especially for growing the inventory of web apps.

It is now possible to use a Linux box for OSP with this release.

This feature is not thoroughly tested, as we will continue using OS X for our lab-based systems. We are using Debian 64-bit for the wearable version of OSP, scheduled for Release 2018c. Please use the Forum to post details if you managed to get Linux release working in a lab setup.