ronitgupta138/Vox-Intel-Engine

ronitgupta138/Vox-Intel-Engine has 45 GitHub stars and 1 forks. Star changes are not measured yet. Last checked 2026-11-08 19:07 UTC. Last push 2026-11-05.

이 수치를 측정하는 방법

🎙️ Real-Time Multi-Modal Speech Telemetry & Acoustic DSP Engine (FastAPI, Python 3.11, Autocorrelation Pitch, FFT Spectral Energy & Lexical Pacing Analytics).

45 스타 1 forks

What you can do

Fundamental Frequency ($F_0$): Normalized autocorrelation over Fourier transform frames estimating vocal pitch register ($75 ext{ Hz} - 500 ext{ Hz}$).
Intonation Variety & Pitch Jitter ($\sigma_{F_0}$): Measures standard deviation around mean pitch to detect flat/monotone delivery vs expressive presentation style.
Root Mean Square (RMS) Volume Dynamics: Decibel-scale loudness measurement ($20 \log_{10}( ext{RMS})$) and peak-to-floor dynamic contrast.
Micro-Pause Segmentation: Automated detection of hesitation gaps ($>300 ext{ms}$) vs natural rhythmic pauses.

측정된 스타 증가

in 24h: 업데이트 지연
in 7 days: 업데이트 지연
in 15 days: 업데이트 지연
in 1 month: 업데이트 지연

서로 겹치는 기간의 순증가량이며 일별 합계가 아닙니다. 측정값이 없으면 비워 둡니다. 스타는 관심도를 나타내며 품질을 보장하지 않습니다.

README.md

Original

██╗   ██╗ ██████╗ ██╗  ██╗  ██╗███╗   ██╗████████╗███████╗██╗     
██║   ██║██╔═══██╗╚██╗██╔╝  ██║████╗  ██║╚══██╔══╝██╔════╝██║     
██║   ██║██║   ██║ ╚███╔╝   ██║██╔██╗ ██║   ██║   █████╗  ██║     
╚██╗ ██╔╝██║   ██║ ██╔██╗   ██║██║╚██╗██║   ██║   ██╔══╝  ██║     
 ╚████╔╝ ╚██████╔╝██╔╝ ██╗  ██║██║ ╚████║   ██║   ███████╗███████╗
  ╚═══╝   ╚═════╝ ╚═╝  ╚═╝  ╚═╝╚═╝  ╚═══╝   ╚═╝   ╚══════╝╚══════╝

VoxIntel Engine — Multi-Modal Speech Telemetry & Acoustic DSP Engine (FastAPI & Python 3.11)

📌 Architectural Overview

VoxIntel Engine is an asynchronous, high-throughput microservice that analyzes technical presentations, executive pitches, and mock interview speeches across multiple parallel acoustic and semantic streams:

                           [ Audio Stream (WAV/PCM) + Transcript ]
                                              │
                     ┌────────────────────────┴────────────────────────┐
                     ▼                                                 ▼
     ┌───────────────────────────────┐                 ┌───────────────────────────────┐
     │   Acoustic DSP Analyzer       │                 │   Lexical NLP Analyzer        │
     │ - FFT Fundamental Pitch (F0)  │                 │ - Words Per Minute (WPM)      │
     │ - RMS Energy & Volume (dB)    │                 │ - Filler Word Density         │
     │ - Silence / Hesitation Pauses │                 │ - Type-Token Ratio (TTR)      │
     └───────────────┬───────────────┘                 └───────────────┬───────────────┘
                     │                                                 │
                     └────────────────────────┬────────────────────────┘
                                              │
                                              ▼
                             ┌─────────────────────────────────┐
                             │   Multi-Modal Fusion Engine     │
                             │ - Calibrated Scoring Matrix     │
                             │ - Delivery Grade (A+ to C)      │
                             │ - Actionable Coaching Insights  │
                             └─────────────────────────────────┘

⚡ Core Telemetry & Mathematical Capabilities

Acoustic Digital Signal Processing:
- Fundamental Frequency ($F_0$): Normalized autocorrelation over Fourier transform frames estimating vocal pitch register ($75 ext{ Hz} - 500 ext{ Hz}$).
- Intonation Variety & Pitch Jitter ($\sigma_{F_0}$): Measures standard deviation around mean pitch to detect flat/monotone delivery vs expressive presentation style.
- Root Mean Square (RMS) Volume Dynamics: Decibel-scale loudness measurement ($20 \log_{10}( ext{RMS})$) and peak-to-floor dynamic contrast.
- Micro-Pause Segmentation: Automated detection of hesitation gaps ($>300 ext{ms}$) vs natural rhythmic pauses.
Lexical & NLP Metrics:
- Speaking Pace (WPM): Classifies cadence against standard presentation benchmarks ($125 - 155 ext{ WPM}$ ideal).
- Filler Word Density: Regex token analysis detecting common hesitation crutches (um, uh, like, actually, basically, you know).
- Lexical Diversity (Type-Token Ratio): Vocabulary richness calculation ($TTR = rac{ ext{Unique Tokens}}{ ext{Total Tokens}}$).
Multi-Modal Confidence Matrix:
- Weighted multi-dimensional scoring evaluating Pacing (25%), Fluency (25%), Vocal Energy (25%), and Intonation (25%) to output an overall calibrated confidence index ($0 - 100$).

📡 REST API Reference

1. Multi-Modal Audio Analysis

POST /api/v1/analyze/audio (Multipart Form)

file: Audio recording (.wav)
transcript (optional): Spoken text transcript

2. Transcript-Only Pacing Analysis

POST /api/v1/analyze/text (JSON Body)

{
  "transcript": "Good morning. In this presentation, I will discuss distributed consensus algorithms and our high-throughput transaction architecture.",
  "duration_seconds": 15.0
}

3. Health & Telemetry

GET /api/v1/health

🚀 Quick Start (Docker)

git clone https://github.com/ronitgupta138/vox-intel-engine.git
cd vox-intel-engine

# Start service container
docker compose up --build -d

# Interactive Swagger UI available at:
# http://localhost:8000/docs

🧪 Running Unit Tests

pytest -v tests/

📜 License

This project is open-source under the MIT License.