Skip to content

Standards

A standard is a named set of checks that decides what "passing" means for a report. Every report — from an upload or from a monitored podcast feed — is judged against one, and the standard used is recorded on the report so editing or replacing it later never changes a report that already exists.

You manage standards from Settings → Standards.

What a standard holds

A standard is a list of checks (loudness, peak level, chapters, and so on — the same checks described in Understanding Reports), and each one is set to:

  • Required — decides the report's Passed/Failed verdict
  • Informational — runs and is shown, flagged when it's outside its limit, but never fails the report
  • Off — doesn't run at all, and isn't charged for

Most checks also carry limits shaping what "within range" means for that standard — a loudness window in LUFS, a maximum peak level, a minimum number of chapters, and so on.

Four checks have no limit to set. They always use the same fixed rule, whichever standard they're in — you can only choose their severity:

  • Quiet channel — fails when one channel is silent while the other carries audio
  • Cover image — fails unless the embedded artwork is 512–3840px on each side and 512KB or smaller
  • Metadata tags — fails unless the file has a title, album and artist tag
  • Explicit flag — fails when swearing heard with 75% confidence or more is found and the episode isn't marked explicit in your feed (it has its own section below)

A standard also decides whether its reports are transcribed — see Transcript below.

Transcript

Every standard has a Transcript option: a full transcript of the episode on the report, with WebVTT, SRT and text downloads.

  • It's on by default. Untick it in your own standard and reports judged by it skip transcription, and aren't charged for it. The report page then says the standard doesn't include a transcript in place of the transcript section.
  • Speech checks need it. Profanities, restarted sentences, the explicit flag, unwanted words and expected words all work from the transcript. While any of them is Required or Informational the option stays ticked and can't be changed, and the editor lists the checks that need it. Turn them all Off and your own choice applies again.
  • It's charged once per report, whether a check needs it or you kept it for its own sake. The editor's estimated cost includes it whenever it runs.
  • No transcript, no automatic chapters from the report. A report without one doesn't offer "Generate automatically" (see Enhancing Audio); you can still type or import chapters.
  • Every built-in standard includes the transcript.

Noise detection follows the same rule with no option of its own: it runs only while the Unwanted noises check is Required or Informational, so turning that check Off skips it, and its cost, too.

Built-in standards

Every workspace can use three read-only standards out of the box:

StandardLoudnessPeak
Podcast (the default)-17 to -15 LUFSat most -1 dBFS
AES speech-19 to -17 LUFSat most -1 dBFS
EBU R128-24 to -22 LUFSat most -1 dBFS
  • Podcast is Audio Audit's own default, close to Apple Podcasts' -16 LUFS loudness target. It's what every show and upload uses until you assign something else.
  • AES speech targets speech-led shows, after the recommendation for online audio streaming and on-demand distribution published by the Audio Engineering Society.
  • EBU R128 is the European broadcast loudness standard, centred on a -23 LUFS target.

All three peak limits above are checked against 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 -1 dBFS here can still overshoot slightly on playback or once it's re-encoded.

Every other check in the built-ins

Apart from loudness, the three built-ins are identical: the same checks, the same severities and the same limits. These are also the starting values when you create a standard of your own.

CheckSeverityLimit
PeakRequiredat most -1 dBFS
Bit rateRequired128 to 256 kbps
Sample rateRequired44,100 to 48,000 Hz
Quiet channelRequiredfixed rule
Explicit flagRequiredfixed rule
Number of channelsInformational2 (stereo) — so a mono file gets a note
Silence at the startInformational500 to 1,000 ms
Silence at the endInformational1,000 to 5,000 ms
Silences in the middleInformationalnone of 3 seconds or longer
Noise floorInformationalat most -60 dB
Sample widthInformational16-bit
ProfanitiesInformationalat most 0
Restarted sentencesInformationalat most 0
Unwanted noisesInformationalat most 0
ChaptersInformationalat least 1
Cover imageInformationalfixed rule
Metadata tagsInformationalfixed rule
RMS loudnessOff-23 to -18 dB when you turn it on
Unwanted wordsOffyour own list
Expected wordsOffyour own list

Unwanted words, expected words and RMS loudness are Off in every built-in — they only make sense once you've told them what to look for or which platform you're targeting, so you turn them on in your own standard. RMS loudness's starting range, -23 to -18 dB, is the one audiobook platforms such as ACX ask for.

Creating your own standard

Settings → Standards lists the Built-in standards and, below them, your workspace's Custom standards. There are two ways to start one:

  • "Add standard", next to the custom standards list. This opens a new standard that starts from the Podcast built-in's checks, severities and limits, ready to edit.
  • Duplicate a built-in. Each built-in's row has a "Duplicate" action, and a built-in opened from the list has a "Duplicate to customise" button. You're asked for a name (a unique one is suggested), then the copy opens ready to edit. This is the way to start from AES speech or EBU R128's numbers.

A custom standard you can edit has no Duplicate action of its own — if you want another copy of it, duplicate a built-in instead and re-apply your changes on top.

The standard editor

The editor has:

  • Name — required, and unique within the workspace (ignoring capitals), up to 100 characters
  • Description — optional, shown under the name in the custom standards list
  • A table of every check, with its Severity (Required, Informational or Off) and its Limit. A limit is a range ("-18 to -16" LUFS), a single "at most" or "at least" value, a choice (sample width, number of channels), a number of seconds for silences in the middle (3 or more), or a word list. The fixed-rule checks show their rule instead. Setting a check to Off greys out its limit.
  • Transcript — whether its reports are transcribed (see Transcript above). It stays ticked while a speech check is Required or Informational.
  • Estimated cost — roughly how many credits a report against this standard costs per hour of audio. It goes down as you turn checks Off, because reports are only charged for the checks their standard runs.

Click "Create standard" (or "Save changes" when editing an existing one) to save.

For example, you might narrow the loudness window to -18 to -16 LUFS, or move Profanities from Informational to Required so any swearing fails the report outright.

Every standard needs at least one Required check — one with nothing Required could never fail, so it can't be saved.

Changes apply to new reports only. Editing a standard — a limit, a severity, its name — never rewrites the verdict of a report that was already judged against it. Every report freezes a copy of the standard it used at creation, so past results stay exactly as they were.

Setting the workspace default

The Default column in both lists shows which standard is the "Workspace default". Click "Set as default" on any other standard — built-in or custom — to make it the default instead. Any show or upload left on "Workspace default" follows it from its next report.

Deleting a standard

Use the delete icon on a custom standard's row, then confirm. Shows using it fall back to the workspace default; if it was the workspace default, the workspace falls back to the built-in Podcast standard. Reports already judged by it keep their result. The Shows using it column tells you how many shows would be affected. Built-in standards can't be deleted.

Plans and limits

Custom standards work on every plan, Free included. A workspace can have up to 10 custom standards of its own — the built-ins don't count towards that. Once you reach 10, "Add standard" and "Duplicate" are disabled until you delete one; editing, assigning and deleting the ones you have still work as normal.

Explicit flag

The explicit flag check (Required in every built-in) is about one specific mismatch: swearing that isn't reflected in your podcast app listing.

  • It fails only when the episode contains swearing and the episode isn't marked explicit in your feed's itunes:explicit tag.
  • An episode with swearing that is marked explicit passes — that's the flag doing its job.
  • An episode with no swearing at all passes regardless of the flag.
  • Only swearing heard with 75% confidence or more counts. The transcription sometimes mishears an ordinary word as a swear word, and one mishearing shouldn't fail a Required check, so a hit below 75% never fails the explicit flag, however many there are. It still appears in the Profanities check, marked "Low confidence — didn't fail the explicit flag", and the explicit-flag check says how many low-confidence hits it ignored.
  • A missing itunes:explicit tag counts as not explicit, the same way podcast apps read it — this check applies to every feed episode, not just ones that carry the tag.
  • It's not applicable, and no row is shown, for uploads (no feed to read a flag from) or for feed episodes whose flag Audio Audit hasn't determined yet — episodes ingested before this check shipped, until a poll refreshes them while they're still the feed's newest item. Either way it doesn't count towards the report's total.

The flag is read live, not just at the time of the report. Each time Audio Audit polls your feed, it re-reads the newest episode's itunes:explicit value — for as long as that episode stays the newest item in the feed. So if an episode fails this check, you go and mark it explicit in your hosting platform (Megaphone, Buzzsprout, and so on), and Audio Audit polls the feed again before your next episode goes out, the report turns green on its own — no retry, no extra credits. Once a newer episode is published, the window closes and that episode's flag is no longer refreshed.

Because a report's verdict can change after it completed this way, an email or webhook already sent for it is never re-sent — but reopening the report itself always shows the current, correct verdict.

Every swearing hit in this check's result is timestamped, exactly like the separate Profanities check, so you can click straight through to each occurrence. Both checks use the same list of swear words, matched on whole words regardless of capitals. Mild words such as "damn" and "hell" aren't on it, and censored transcriptions such as "f**k" still count.

Unwanted words and expected words

These are two checks you configure yourself with your own word or phrase lists — both Off in every built-in standard, so you only see them once your own standard turns them on. In the editor each has a box where you type one word or phrase per line.

Unwanted words fails when any of your listed words or phrases are said — a discontinued slogan, a competitor's name, anything you don't want in the episode. Every occurrence is timestamped so you can jump straight to it.

Expected words is the mirror image: it fails when any listed phrase is never said at all — a sponsor read, a membership pitch, a legal disclaimer. The result shows every phrase you listed, with timestamps for the ones that were found and a clear "not found" for the ones that weren't.

Both checks share the same list limits: up to 50 words or phrases, each up to 50 characters, matched case-insensitively on whole words (so "Ford" won't match inside "afford"). Blank lines and repeats are dropped when you save. A check you've turned on (Required or Informational) needs at least one word or phrase in its list.

Silences in the middle

Separately from the silence checks at the start and end of an episode, Audio Audit flags any silence inside the body of the episode that's the set length or longer — 3 seconds by default, which is also the shortest limit you can set. It's Informational in every built-in standard, so a long mid-episode silence is shown but doesn't fail the report unless you move it to Required in your own standard.

Assigning standards

A standard has to be assigned somewhere before it judges anything:

  • A show — on that show's settings page (see Podcast Details), pick "Workspace default" or a specific standard from the Standard picker. Every feed report for that show is judged against it.
  • An upload — pick a standard from the Check against control at the top of the Dashboard's Generate Report card before you drop the file (see Creating Reports). Leave it on the default and the workspace's default standard applies.
  • A new podcast — the same Check against control applies to a podcast you add from the Dashboard's search box: if the show is new to your workspace, it keeps that standard for its automatic checks. For a show you already have, the choice applies to that one report only and never changes the show's own setting. Owners and Admins can choose Create custom standard… from the same control (while the workspace has fewer than 10 custom standards) and come straight back with the new standard selected.
  • The workspace default — set with "Set as default" in Settings → Standards. Any show or upload left on "Workspace default" follows whatever you set here; leave it unset and Audio Audit's own Podcast standard applies.

Standards belong to a workspace. To work on another Organisation's standards, switch workspace with the switcher at the top of the workspace section in Settings, or the one next to the Audio Audit logo.

Who can do what

ActionWho
Create, edit, duplicate and delete standards; set the workspace defaultWorkspace Owners and Admins
Choose which standard a show usesthat show's Managers, and Owners/Admins
Choose a standard for an uploadanyone with the "Run reports" workspace permission (see Roles and Permissions)

Other members who open Settings → Standards see a "Permission required" notice instead of the lists. In a personal workspace, you do all of the above yourself.

Next steps