speech-evaluation

Versatile Evaluation of Speech and Audio

Evidence-aware Japanese speech visualization: pitch and intonation, waveform inspection, and experimental alignment.

Comprehensive voice cloning evaluation: seven automatic speech metrics and ready-to-use benchmark datasets. Score any model’s audio with Python or CLI.

A simple pronunciation evaluation engine python library.

An objective evaluation toolkit for multi-speaker dialogue generation using speaker attribution, speaker similarity, and word error rate

Mandarin Tone Classifier

SHALLOW, the first hallucination benchmark for ASR models

Created a Speech Quality Evaluator API 🎤 as part of an educational technology project, assessing accuracy, pronunciation, and vocal expression in real-time. 📈 Designed to provide weighted average scores for speech evaluations, ensuring precise feedback for educational purposes. 🌟

A standalone tool for evaluating Automatic Speech Recognition (ASR) models, particularly optimized for medical/clinical speech recognition, using Word Error Rate (WER) metric

Evaluation of speech AI behaviour, including transcription, speech metrics, filler detection, and grounded coaching design.

Blind listening test comparing human and AI-generated speech in Brazilian Portuguese.

Self-hosted and offline-first platform for subjective speech evaluation (MOS & SUS intelligibility testing).

Measured Bangla text-to-speech benchmark and a Bangla LoRA for an open TTS model

Kyrgyz-language AI Tutor portfolio: annotation, transcription benchmark, decision log, localization audit, TTS scorecard — built on kgvoice

Machine learning notebook for SHL Grammar Scoring Challenge using audio feature extraction and regression models to predict grammar scores.

Reproducible speech evaluation platform for TTS benchmarking, regression detection, and traceable evidence.

🎥🎵 英语音视频复述练习评判系统 — 基于 Gradio、OpenAI Whisper 语音转录与本地 Ollama (Gemma 2) 的英语听说复述 AI 评测系统

Speech recognition benchmark and ASR leaderboard for vertical domain evaluation.

Multilingual speech evaluation benchmarks for ASR and related speech tasks with reproducible dataset preparation and reporting.

An experimental, fully automated TTS evaluation harness built on open models