RecoMadeEasy ® Recognition Technologies Android Speech Transcription and Speaker Diarization
This is a demonstration of the RecoMadeEasy ® Speech and Speaker Diarization of Recognition Technologies, Inc. This demo is running entirely on Android, specifically on a Samsung S6 phone without the aid of any server or cloud service. The speech transcription supports over 300,000 unique words and the speaker identification can do thousands of models in realtime while the speech transcription is being done, solely on the Android device. The engine is as accurate as a full-fledged cloud based or server based engine. It does full diarization on the device by doing speaker segmentation, speaker identification, gender identification, and speech transcription, all at the same time. We also demonstrate the process of speaker enrollment and unenrollment. This system has been developed over the past 20 years and it is entirely written in C++. It is capable of running as standalone on Android, as seen here, as well as client/server configurations with servers distributed over any number of networks. Our website is https://www.recotechnologies.com
Enabling your Android applications to use our speech and speaker diarization, as well as all other modalities described below is very easy. You simply link your Android app with our RecoMadeEasy ® library and install our Android Service. Then when your app starts, our service gets called automatically and will process the audio. You have a choice of either using our library to do the audio capture or provide the audio stream to our library through your own means. Most of our clients are able to do this in a matter of hours, using our sample code.
Please see the other capabilities of our engine which includes face recognition, object detection and recognition, emotion detection, and many other things. In addition to English, the transcription can run with Mandarin Chines, Arabic, Spanish, and German, as well as bilingual combination of these languages.