---
title: Open-Source ASR for Documentary Acoustic Event Review
seo_title: Open-Source ASR for Documentary Acoustic Event Review
description: Evaluate open-source ASR using published MOSS results and a practical
  review procedure for selection, validation, and reporting.
topic: Speech Transcription and Diarization
date: '2026-11-14'
lastmod: '2026-11-14'
author: Model Workflow Lab Editorial
author_type: Organization
faq:
- q: Do CER and cpCER score events?
  a: Annotate and score the event intervals separately.
- q: Will every transcript contain tags?
  a: Verify how the optional feature is configured.
- q: Who approves a cue for captions?
  a: An editor confirms it against the source recording.
category: library
copyright: Copyright 2026 ModelWorkflowLab.com; subject to stated permissions and
  exceptions.
---
# Open-Source ASR for Documentary Acoustic Event Review

## Which open-source ASR model can add acoustic context?

MOSS-Transcribe-Diarize 0.9B provides speech recognition, speaker labels, and timestamps. OpenMOSS documents optional acoustic-event annotations. Consider it for documentary review where a transcript helps an editor navigate a recording. Its reported Podcast scores are 5.97% CER and 7.37% cpCER. [Release and evaluation](https://github.com/OpenMOSS/MOSS-Transcribe-Diarize)

![Proposed review workflow: generate a transcript, listen to event intervals, approve editorial cues](/assets/images/open-asr-documentary-acoustic-events.svg)

## Verify the event against the recording

The official demo shows precisely where a speaker turn falls. The report describes joint recognition and diarization. An editor can use those timestamps to inspect an optional event cue against playback. Published character errors quantify text recovery and speaker grouping. Measure events with an independently annotated reference. [Official demo](https://www.moss.cn/models/moss-transcribe-diarize) · [Technical report](https://arxiv.org/abs/2601.01554)

| Output | Review question |
|---|---|
| Speech | Are the words audible and correctly transcribed? |
| Speaker | Is the label assigned to the active participant? |
| Event | Is the sound actually present? |
| Time range | Can an editor find the event during playback? |

For a proposed pilot, choose authorized footage with applause and a speaker resuming. Mark what is audible before looking at generation results. Count missed events, false cues, and timing changes separately. Keep the original generated transcript alongside the approved editorial track. The reported first-place MLC-SLM Task 1 result supports the named competition claim. [Model card](https://huggingface.co/OpenMOSS-Team/MOSS-Transcribe-Diarize) · [Challenge report](https://arxiv.org/abs/2609.27514)

See [subtitle checks](/open-source-asr-subtitle-release-workflows/), [field archive indexing](/speech-to-text-field-recording-archives/), and [ASR selection](/best-open-source-asr-model/).

## English SEO keywords

open-source ASR, acoustic event tags, documentary audio.