zombie4xx / optimization_report.md
memevis's picture
Upload folder using huggingface_hub
14a750e verified

TTS Model Optimization Report

Overview

  • Optimization Date: 2025-05-20 16:21:58
  • Iterations Performed: 3
  • Total Strategies Tested: 36
  • Best Overall Score: 0.8718

Performance Metrics

  • Mel Cepstral Distortion: 0.9757
  • Word Error Rate: 0.0580
  • Naturalness: 0.9479
  • Intelligibility: 0.9896
  • Speaker Similarity: 0.9366
  • Prosody: 0.9060
  • Overall Quality: 0.9279

Optimization Insights

Most Effective Parameter Settings

  • attention_scale: 1.4598
  • output_scale: 1.4200
  • projection_scale: 1.4402
  • encoder_scale: 1.2803
  • decoder_scale: 1.3598
  • base_enhancement: 0.0022
  • importance_factor: 1.6635

Optimization Journey

Iteration 1: Best strategy 'balanced_enhancement_var1.5' - Score 0.8718

Iteration 2: Best strategy 'balanced_enhancement_var1.5_iter2_clarity_boost' - Score 0.8624

Iteration 3: Best strategy 'balanced_enhancement_var1.5_iter3_noise_scale_1.2' - Score 0.8637