zombie509 / optimization_report.md
memevis's picture
Upload folder using huggingface_hub
3564768 verified

TTS Model Optimization Report

Overview

  • Optimization Date: 2025-04-27 00:24:51
  • Iterations Performed: 3
  • Total Strategies Tested: 32
  • Best Overall Score: 0.8707

Performance Metrics

  • Mel Cepstral Distortion: 0.9231
  • Word Error Rate: 0.0726
  • Naturalness: 0.9232
  • Intelligibility: 0.9787
  • Speaker Similarity: 0.9678
  • Prosody: 0.9482
  • Overall Quality: 0.9411

Optimization Insights

Most Effective Parameter Settings

  • attention_scale: 1.2996
  • output_scale: 1.3982
  • projection_scale: 1.6402
  • encoder_scale: 1.1996
  • decoder_scale: 1.4243
  • base_enhancement: 0.0017
  • importance_factor: 1.8605

Optimization Journey

Iteration 1: Best strategy 'speaker_similarity_focus_var0.75' - Score 0.8684

Iteration 2: Best strategy 'speaker_similarity_focus_var0.75_iter2_decoder_boost' - Score 0.8707

Iteration 3: Best strategy 'ensemble_top3_iter3' - Score 0.8636