PRODUCTS
-
Emotion Detection from Speech
-
Applications​​
​​
-
Sentiment analysis of clients for any industry
-
Emotion detection for agent and client for all types of calls
-
Emotional state analysis of a student during oral examination
-
Agent performance and customer satisfaction (CSat) analysis for promotional calls
-
Sentiment analysis for speakers in online meetings
-
Segregation of clients based on their readiness
-
Sentiment analysis for stress mode in forensic applications
-
Speaker Diarization
-
Application​
​​
-
Speaker segregation for mono audio file recordings.
-
Speaker segregation for forensic applications from suspect audio.
-
Automatic detection of number of speakers in a conference call using unsupervised system.
-
Diarization and transcription of conference call for speaker audios.
-
Useful for writing minutes of the meeting
-
Forensic Speaker identification
-
Applications​​
​​
-
Recognizing and verifying suspect person in forensic application
-
Speaker verification for entry in restricted area
-
Speaker verification for bank transactions based on speech
-
Speaker identification as one of the culprits from stored database
-
Speaker verification in case of deep fake generated audio
-
Call Audit and Speech Analytics
-
Application
​​
-
Useful for call centres
-
Agent performance and score can be found
-
Customer sentiments can be captured
-
Text obtained using speech to text can be used for text processing to do call audits
-
Intelligent algorithm can help actual % of audit
-
Keyword spotting can be used if speech to text is to be eliminated
-
Fake Audio Detection
-
Application​
​​
-
Check if any tampering is done by suspect on audio files
-
Check if any phase shift is introduced, or pause is introduced in recording
-
Check if 2 audios recorded on different devices are mixed up
-
Check if any dc shift is introduced
-
Check if original sampling frequency is changed
-
Check the metadata file for the audio
-
Check if a true recording or deep fake audio
-
Speech to Text system
-
Convert audio to text. It gives text with punctuation.
-
Use speaker diarization to separate two speaker’s talks.
-
Label agent and client for converted text.
-
Process text as per requirement to generate result in
excel sheet. It may include identifying certain phrases
in text.
-
Batch processing is allowed. System uses multi tasking
environment
-
Text independent, Language Independent Speaker Identification
-
Developed a system that extracts speech signature or
voice print for suspect audio. -
Extracts VAD segments for audio and extracts speaker
embeddings. -
Verifies a speaker if the suspect audio matches with the
control audio. -
The results obtained excellent