MixedVoices

The open-source analytics and evaluation platform for voice agents.
Analyze, visualize, evaluate, and optimize your conversational AI performance.

Core Features

Powerful tools to analyze, evaluate, and optimize your voice agent's performance

Flow Visualization
Interactive flowcharts showing conversation paths and patterns
ML Performance Metrics
Track hallucinations, scheduling accuracy, conciseness, and empathy scores
Call Quality Analysis
Monitor latency, interruptions, and signal quality metrics
Version Control
Track and compare agent behavior across different iterations
Agent Evaluation
Test performance through simulated scenarios and stress testing
Success Rate Analytics
Measure and optimize performance metrics

Call Flow Analysis

Visualize customer journeys and identify optimization opportunities automatically

Automatic Flow Generation

Auto generates interactive call flow visualizations from your conversation data, understand common paths and user behaviors.

Drop-off Analysis

Quickly identify where users dropping or experience friction in their journey with clear termination metrics at each step.

Journey Optimization

Use flow metrics to optimize conversation paths and improve completion rates across different interaction types.

Call flow visualization showing conversation paths

Voice Analytics Suite

Comprehensive metrics to evaluate both technical performance and AI capabilities

Call Quality Metrics

Interruptions
Track and minimize conversation interruptions
Latency
Measure and optimize response delays
Signal Quality
Monitor audio clarity and stability
Words per Minute
Track conversation pace and fluency

ML Performance Metrics

Hallucinations
Track and reduce AI hallucinations
Call Scheduling
Measure appointment scheduling accuracy
Conciseness
Evaluate response clarity and brevity
Empathy
Assess emotional intelligence in responses
Objection Handling
Track successful resolution of concerns
Context Awareness
Evaluate conversation context retention
Adaptive Questioning
Measure dynamic response adaptation
Verbatim Repetition
Identify unnecessary information repetition

Agent Evaluation

Comprehensive testing and validation of your voice agent's performance

Scenario Simulation

Test agent performance using real-world scenarios based on historical recordings

ML Performance

Track key AI metrics including hallucination rates, response accuracy, and emotional intelligence

Stress Testing

Validate performance under various conditions including edge cases and high load scenarios

Agent evaluation dashboard showing performance metrics

Easy Integration

Seamlessly connect with your existing voice agent tools and workflows

Vapi Logo

Vapi

Pipecat Logo

Pipecat

Retell Logo

Retell

Get Started Today

Add MixedVoices's Voice Analysis to your voice agent in minutes

Installation
pip install mixedvoices
Usage
import mixedvoices as mv

# Create a new project
project = mv.create_project("clinic")

# Create a new version
v1 = project.create_version("v1")

# Add recording to analyze
v1.add_recording("call.mp3")

# Run evaluation
evaluator = v1.create_evaluator()
evaluator.run(agent=MyAgent)