Skip to content

Understanding Reports

Audio Audit reports provide comprehensive analysis of your audio files. Learn how to interpret the results and use them to improve your podcast quality.

Report Overview

Each report analyses your audio against the checks in a standard and gives you an at-a-glance verdict plus a detailed breakdown. Access reports from:

  • Dashboard under "Latest Reports" section
  • Reports page showing all your recent reports
  • Podcasts page showing all podcasts within the current workspace — the scores under "Last 5 scores" are clickable
  • Direct link from processing completion email

Audio Player with Visual Markers

The interactive audio player is at the top of each report.

Player Features

  • Waveform visualisation: See audio levels throughout the episode
  • Timestamps: Check episode length and play position
  • Visual markers: Clickable indicators showing:
    • 🟢 Green markers: Chapters and sections
    • 🟡 Amber markers: Timestamped occurrences (swear words, unwanted words, unwanted noises, long silences, and expected phrases where they were found) — a missing expected phrase has no timestamp to mark, so it only appears in the check itself, not on the waveform
  • Playback controls: Play/pause, scrubbing, back/forward 10 seconds at a time
  • Timeline scrubbing: Click anywhere on the slider to jump to that position

Using Markers

  1. Hover over markers to see descriptions
  2. Click markers to jump directly to that timestamp
  3. Use markers to quickly review flagged issues
  4. Markers correspond to items in the checks section

Standard and Verdict

Every report is judged against a standard — a named set of checks with their own limits. The report header shows "Standard: …" naming the one that applied, and every workspace starts on Podcast, Audio Audit's own default. See Standards for the built-in standards and for creating your own.

Each check in a standard is one of three things:

  • Required: decides whether the report passes. A single failed Required check fails the whole report.
  • Informational: still runs and still shows its value, flagged when it's outside the standard's limit, but never affects the Passed/Failed verdict.
  • Off: doesn't run for this standard at all, and isn't shown.

Passed or Failed

Every completed report shows a Passed or Failed verdict, decided purely by the Required checks: Passed means every Required check came back within its limit; Failed means at least one didn't. On the report page the score card leads with the word Passed or Failed, and a Failed report also shows how many Required checks failed. Report lists — on the Dashboard, the Reports page and each podcast's page — show the verdict as a green tick (Passed) or red cross (Failed) beside the score; hover over it to see the word.

This is a separate signal from the percentage, and the two can read differently on the same report:

  • A report can read 94% and still be Failed, if the one thing that broke was a Required check such as loudness or peak.
  • A report can read 60% and still be Passed, if everything that failed was Informational only — several notes about silences or a missing chapter marker, say, with nothing Required broken.

Read the Passed/Failed verdict for "does this meet the standard", and the percentage for "how much room is there to improve".

Score Breakdown

  • Header: the verdict and the percentage side by side, e.g. Failed and 58%
  • Footer: "10 / 17 checks passed" — checks that met the standard's limit, out of the total — plus "1 required failed" on a Failed report
  • Visual bar: green for passes; the non-passing share splits into red (Required failures) and amber (everything else — Informational checks outside their limit)

While a report is still processing, the card shows just the running count, such as "10 / 17", and the percentage.

The percentage still counts every applicable check equally, Required and Informational together, exactly as it always has — it is not a Required-only score.

Required and Informational Checks

Each result row in the report shows:

  • A status pill: FAIL (red) or PASS (emerald) for a Required check; NOTE (amber) or OK (grey) for an Informational one outside or within its limit
  • The measured value
  • A limit line underneath, such as "Standard: -17 to -15 LUFS" — this comes from the standard assigned to the report, not a fixed rule, so it changes if a different standard is assigned

Required checks are listed first, then Informational checks in their own section, which starts collapsed when everything in it passes.

The exact checks that run, and their limits, depend on which standard judged the report — see Standards for the built-in Podcast, AES speech and EBU R128 standards and their numbers, and for creating and customising your own.

What the checks look at

Technical measurements — bit rate, sample rate, average loudness (LUFS), peak level, noise floor, sample width, channel count (mono/stereo) and quiet-channel detection (catching a channel that's gone silent, such as a dead mic, separately from the channel-count check).

Peak is measured as sample peak — the loudest individual sample in the file — not true peak (the level the waveform can briefly reach between samples once it's reconstructed for playback, measured by oversampling). A file that just clears its peak limit in Audio Audit can still overshoot slightly on playback or once it's re-encoded.

Content checks — silence at the start, end, and anywhere else in the episode; chapter markers; unwanted noises such as coughs, sneezes or pets; cover artwork; and metadata completeness (title, artist, album).

Speech checks — profanities and restarted sentences, plus whether swearing episodes are correctly marked explicit in your feed. That explicit-flag check applies to every feed episode — a feed with no itunes:explicit tag at all counts as not explicit, the same way podcast apps read it — and only skips episodes on uploads (no feed to read) or whose flag isn't known yet (see Standards for how that check works). A custom standard can also add unwanted-words and expected-words checks of your own — alerting when a phrase is said, or missing.

Automated Transcription

Every report is transcribed unless it was judged by one of your own standards with its Transcript option switched off; such a report says so in place of the transcript.

New reports are transcribed with OpenAI's Whisper large-v3-turbo model running on GPU; reports made before this upgrade keep the transcript they were originally created with.

Transcription Features

  • Full text: Complete speech-to-text conversion
  • Timestamps: Clickable times to jump to audio
  • Formatting: Paragraph breaks for readability
  • Speaker labels: When more than one voice is detected, labels such as "Speaker 1" and "Speaker 2" mark each change of speaker in the transcript, and every cue in the downloads names its speaker
  • Copy and download: "Copy to clipboard", or Download the transcript as WebVTT (transcript.vtt), SRT (transcript.srt) or plain text (transcript.txt)

Speaker detection uses pyannote.audio's community-1 pipeline (CC BY 4.0). Speakers are numbered in the order they first speak. Labels only appear on reports made after this feature launched, and only when more than one speaker is found — a single-voice episode, or one where detection couldn't identify speakers, shows the transcript without labels.

Using Transcriptions

  1. SEO benefits: Search engines can index your content
  2. Accessibility: Makes content available to deaf/hard-of-hearing audiences
  3. Content repurposing: Create show notes or blog posts
  4. Quote extraction: Find exact quotes with timestamps

Transcription Accuracy

  • Most accurate on clear, well-recorded speech
  • Always proof-read before publishing — it may struggle with:
    • Heavy accents
    • Technical jargon
    • Multiple overlapping speakers
    • Poor audio quality

Sharing Reports

Audio Audit reports can be shared with others.

Sharing Options

  • Direct link: Click Share under the report title for a "Share report" window with the report's link, a copy button and a QR code
  • Team members: Create an organisation and invite team members

Anyone with a report's link can open it — they don't need an Audio Audit account — so share it only with people you're happy to show the results to.

Use Cases

  • Share with podcast editors for improvements
  • Show sponsors your quality standards
  • Collaborate with team members
  • Document quality improvements over time

Taking Action on Results

Prioritising Improvements

Start with anything under Required checks — those are what decide the Passed/Failed verdict, so fixing them is what turns a Failed report into a Passed one.

Then work through Informational notes as time allows — silences, missing chapters, unwanted noises and the like. They don't fail the report, but they still count towards the percentage and are worth cleaning up for listener experience.

Common Fixes

You can find more info about a check by clicking it to expand it. From there you may also see a "More info" link that opens more help in a popup window. Our Podcasting Articles includes several tutorials for specific DAWs.

Tracking Improvements

On the Dashboard and Podcasts page you can see a graph of scores over time for each podcast in your current workspace.

Report History

  • Compare scores over time
  • Track specific measurement improvements
  • Document your quality journey

Best Practices

  1. Analyse every episode before publishing
  2. Keep a log of changes made
  3. Set quality targets (e.g., 85% minimum score)
  4. Review trends monthly

Getting Help

Understanding Specific Checks

  • Expand a check and click "More info" for a detailed explanation
  • Check FAQs for common questions
  • Contact support for clarification

Technical Assistance

  • Report ID helps support assist you
  • Screenshots of issues speed up resolution
  • Include your workflow details when asking for help

Next Steps