wer

Evaluate your speech-to-text system with similarity measures such as word error rate (WER)

MeetEval - A meeting transcription evaluation toolkit

한국어 STT 출력의 CER, WER, CRR, 키워드·개체명·코퍼스 평가를 제공하는 Python 패키지

🐍📦 Ultra-fast Python package for calculating and analyzing the Word Error Rate (WER). Built for the scalable evaluation of speech and transcription accuracy.

A lightweight library for normalizing speech transcripts before computing WER

Calculates the word error rate of two strings, and the result is written into beautify HTML.

Machine Translation (MT) Evaluation Scripts

🐍📦 Easy-to-use Python package for lightning-fast Word Error Rate (WER) analysis

pytest for voice agents — open-source testing and observability for voice AI. Simulate real callers over real audio, measure exact WER against known ground truth, and attribute every failure to STT, LLM, or TTS. Self-hosted, Apache-2.0.

A simple Python package to calculate word error rate (WER).

WER распознавания русской речи по трём доменам: Whisper large-v3-turbo, GigaAM v3, Deepgram Nova-2/Nova-3. Ранжирование движков переворачивается между доменами; сырые транскрипты опубликованы

Exploring the functionality of the werpy Python package through testing within the Gradio tool for interactive user interface development.

ASR scoring and annotation QA toolkit for researchers

WER and inference-cost benchmark for Indian-English ASR, and an audit of the evaluation itself: 9 systems x 3 corpora, cluster-bootstrap CIs with Holm correction, a reference-artifact taxonomy confirmed by human re-transcription, and a speaker-disjoint fine-tuning study across 6 seeds x 3 sizes.

Python implementation of the metric Paraboth with some improvements, such as also workling on Corpus and using a LLM to paraphrase.

Research dashboard & evaluation pipeline for compact Whisper models (Tiny/Base/Small) on Urdu-English code-mixed speech. FastAPI dashboard + SQLAlchemy backend analyzing WER/SPER across 327 recordings from 11 bilingual speakers, in a zero-shot setting.

Reproducible Whisper fine-tuning, distillation, ACFT, and whisper.cpp conversion pipeline for Hebrew speech recognition.

Quality Gate — CER/WER/medical-term accuracy with nightly regression checks

Reproducible speaker diarization (DER) and German speech recognition (WER) benchmarks — re-score pyannote-community-1 on VoxConverse, CALLHOME-de and CommonVoice with jiwer and pyannote.metrics, no GPU needed.

Lightweight ASR evaluation harness for measuring Word Error Rate, character accuracy, and inference latency across Whisper and other speech recognition models.