{% extends "base.html" %} {% block title %}{{ recording.title }} — Evaluation{% endblock %} {% block page_header %} {% include "partials/recording_header.html" %} {% endblock %} {% block content %} Reference Your transcription {{ eval_ctx.human_chunks|default(0) }}/{{ eval_ctx.total_chunks|default(0) }} chunks · {{ eval_ctx.human_words|default(0) }} words Edit in Transcribe → compared to Hypothesis LLM transcript {{ eval_ctx.llm_chunks|default(0) }}/{{ eval_ctx.total_chunks|default(0) }} chunks matched {% if eval_ctx and eval_ctx.has_llm_source %} Loaded{% if eval_ctx.llm_format %} · {{ eval_ctx.llm_format }}{% endif %} {% else %} Add below ↓ {% endif %} Add LLM transcript Paste or upload the model output. We auto-detect format and match lines to your audio chunks by time. Paste Upload file Auto-detect English Apply Auto-detect English Upload LLM status Remove LLM text —Chunks —With LLM —Coverage —Format Run evaluation WER scores: semantic (fair) vs strict (exact words) Evaluate Results Semantic WER — Strict WER — —Semantic accuracy —Strict accuracy —Quality Errors Substitutions0 Deletions0 Insertions0 Per-chunk comparison {% endblock %} {% block scripts %} {% endblock %}
Paste or upload the model output. We auto-detect format and match lines to your audio chunks by time.
WER scores: semantic (fair) vs strict (exact words)