Skip to content

Releases: bunyaminergen/Callytics

v1.1.0 — MP-SENet Speech Enhancement Integration

01 Jan 17:27
@bunyaminergen bunyaminergen
v1.1.0
de8a874
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Learn about vigilant mode.
Compare
Choose a tag to compare
v1.1.0 — MP-SENet Speech Enhancement Integration Latest
Latest

Release Title
v1.1.0 — MP-SENet Speech Enhancement Integration

What's New

  • MP-SENet Integration: Added an advanced deep learning–based speech enhancement module using MP-SENet Paper.
  • Configuration Updates: Updated config.yaml to include models.mpsenet.model_name, allowing easy switching or upgrading of the MP-SENet model.
  • Sample Data: Included a sample noisy audio file for demonstration and testing.
  • Documentation: Revised documentation/instructions to guide you through enabling MP-SENet–based enhancement in your existing pipeline.

Highlights

  • Backward compatible: Existing Denoiser–based workflows remain unchanged unless you explicitly replace them with the new SpeechEnhancement steps.
  • Easier configuration: Simply update models.mpsenet.model_name in config.yaml to point to the desired model on Hugging Face.
  • Enhanced audio quality: Achieve better noise reduction and speech clarity thanks to the MP-SENet architecture.

Upgrade Notes

  • This release should not introduce any breaking changes for existing workflows.
  • If you do not wish to use MP-SENet, simply continue using the Denoiser class as before.

Thank you for trying out MP-SENet with Callytics! Please open an issue if you experience any problems or have suggestions for improvement.

Assets 2
Loading

v1.0.0 Initial

22 Dec 17:55
@bunyaminergen bunyaminergen
v1.0.0
aade410
Compare
Choose a tag to compare
v1.0.0 Initial

v1.0.0 – Initial Version

Overview

This marks the first official release of Callytics, focusing on call analytics with speech recognition and LLM-driven NLP.

New Features

  • Sentiment Analysis
  • Profanity Word Detection
  • Summary Generation
  • Conflict Detection
  • Topic Detection

Improvements & Changes

  • Automated pipeline triggered by audio files in .data/input.
  • Database integration for storing analysis results.
  • Enhanced documentation (architecture, setup, mathematical foundations).

Known Issues

  • High GPU requirements for large LLMs.
  • Code adjustments needed if you change the database schema.
  • No real-time (live stream) support yet.

Breaking Changes

  • None for this initial stable release.

Future Plans

  • Speech Emotion Recognition
  • Forced Alignment Model (Transformer-based)
  • Vocal Separation Model
  • Comprehensive Unit Tests
  • Real-Time Call Analysis

Note: Please report any issues or suggestions through GitHub Issues.

— Callytics Team

Full Changelog: https://github.com/bunyaminergen/Callytics/commits/v1.0.0

Assets 2
Loading