> ## Documentation Index > Fetch the complete documentation index at: https://docs.premierecopilot.com/llms.txt > Use this file to discover all available pages before exploring further. # Smart Profanity > Find swear words, violence and everything video platforms flag, then bleep or mute only the voice. Music and picture stay untouched. ![Smart Profanity bleeping the swear words of an interview in the timeline](https://mintcdn.com/premieregpt/QduopnJH0A5j0meC/images/smartprofanity-hero.jpg?fit=max\&auto=format\&n=QduopnJH0A5j0meC\&q=85\&s=8fa7f5508fd72561e867dd55486089dd) Smart Profanity transcribes your sequence, finds the words that video platforms and social networks flag (swearing, insults, sexual terms, violence, self-harm, hate, hard drugs), and bleeps or mutes them on the **voice track only**. Music, sound effects and picture keep playing. Every word is judged **in its context**: "kill time", "you killed it" or a first name that looks like a swear word are never bleeped. You review everything before anything changes on the timeline. Works in **Premiere Pro** and **DaVinci Resolve**. ## Quick Start **1. Open Smart Profanity** Open the **Tools Menu** and select **Smart Profanity**. **2. Choose the range** Select **Whole timeline**, or **In / Out** to check only the range between your In and Out points. **3. Set the detection** Choose a level (**Light**, **Standard** or **Strict**) and the categories to bleep. **4. Choose the bleep** **Bleep**, **Mute** or **Markers only**. For a bleep, pick a sound or import your own, and set its volume. **5. Analyze** Click **Find the words to bleep**. The arrow on its left picks the transcription engine. You watch every sentence being checked and every flagged word turning red, one by one. **6. Review and apply** Uncheck what should stay audible, listen to any word with its bleep, search the transcript for missed words, then click **Bleep N words**. > [!WARNING] > > Keep Premiere Pro or DaVinci Resolve open and do not edit the timeline while Smart Profanity is analyzing or applying. ## Choose the range | Range | What it does | | - | - | | **Whole timeline** | Checks the sequence from beginning to end. | | **In / Out** | Checks, charges and bleeps only between the sequence In and Out points. | ## Settings The settings are grouped in three blocks: **Detection**, **Bleep** and **Advanced settings**. ![Smart Profanity Detection and Bleep settings](https://mintcdn.com/premieregpt/QduopnJH0A5j0meC/images/smartprofanity-settings.png?fit=max\&auto=format\&n=QduopnJH0A5j0meC\&q=85\&s=f3bb2d7e4377dd308f74054727541f05) ### Language Leave **Auto** to detect the language. Set it yourself when the video mixes several languages or when detection is wrong. Detection works in every supported language, including Japanese, Korean and Hindi. ### Level | Level | What gets bleeped | | - | - | | **Light** | Only what is clearly offensive. | | **Standard** | What video platforms and advertisers usually flag. Recommended. | | **Strict** | Also borderline words. Use it for kids or brand content. | Words that sit just below the chosen level appear in the review under **To check**, highlighted in yellow and unchecked. One click bleeps them. ### What to bleep Six categories, all enabled by default: | Category | Examples | | - | - | | **Swearing** | Swear words, vulgar insults and offensive expressions. A whole expression ("son of a bitch") is bleeped as one block. | | **Sexual** | Explicit sexual terms. | | **Violence** | Words describing or threatening real violence ("I'll kill you", "murder", "stabbed"). | | **Self-harm** | Suicide, self-harm, telling someone to kill themselves. | | **Hate** | Slurs against a group of people. | | **Drugs** | Hard drugs and their use. | > [!NOTE] > > Gameplay is not violence: "I killed the boss" in a gaming video is kept. Idioms and harmless look-alikes are kept too. ### Treatment | Treatment | What happens | | - | - | | **Bleep** | The voice is muted under each word and a bleep plays over it, on its own track. | | **Mute** | The voice is muted under each word, nothing plays over it. | | **Markers only** | A marker is added on each word. The audio does not change. | ### Bleep sound Choose **Classic bleep**, **High bleep**, **Retro** or **Cartoon**, or click **Your sound** to import any audio file. Every bleep lasts **exactly as long as the word**, with short fades so it never clicks. An imported sound is looped or trimmed to fit. **Bleep volume** goes from -20 to +6 dB. What you hear with the play button is exactly what lands on the timeline. ### Advanced settings | Setting | Default | What it does | | - | - | - | | **Voice tracks** | Tracks with clips | Only the checked tracks are muted under each bleep. Tracks whose name suggests music or sound effects are left out: they keep playing under the bleep. | | **Margin around each word** | 40 ms | Extra time bleeped before and after each word, so no letter slips through. | | **Keep the first syllable** | Off | The beginning of the word stays audible ("f-BLEEP"), as on TV. | | **Add a marker on each word** | On | A marker named with the masked word ("f\*\*\*\*\*g") on every bleep. | | **Always bleep** | Empty | Words or phrases to bleep every time, separated by commas. `*` is a wildcard: `frick*` also catches "fricking". The paste button pastes a whole list at once. | | **Never bleep** | Empty | Words that must stay audible even if they are flagged. | > [!WARNING] > > A locked track cannot be muted. Smart Profanity warns you before the analysis when a checked voice track is locked. Your last settings are remembered. Save named presets with the preset menu. ## Transcription engine A word has to be heard to be bleeped, so the engine matters. Click the arrow on the left of **Find the words to bleep** to pick it. **PremiereGPT** is selected by default, and the button shows the total price of the run. ![Transcription engine menu of Smart Profanity](https://mintcdn.com/premieregpt/QduopnJH0A5j0meC/images/smartprofanity-engine.png?fit=max\&auto=format\&n=QduopnJH0A5j0meC\&q=85\&s=6b90d70636379e033ae2341436a5f98b) | Engine | Best for | | - | - | | **PremiereGPT** | The fastest and cheapest. Good for clear speech. | | **PremiereGPT Large** | More accurate on accents, fast speech and slang, a bit slower. | | **PremiereGPT V2** | Transcribes word for word and never softens a swear word. The most reliable choice for heavy or mixed-language content. | > [!TIP] > > When the transcript already hides a word ("f\*\*\*" or "\*\*\*"), that word is bleeped automatically, whatever the engine. Every transcription made by Smart Profanity is added to the **Recent transcriptions** of Smart Subtitles, so you can caption the same sequence without transcribing it again. ## Watch the analysis While Smart Profanity works, the progress window shows the algorithm live: 1. **Sentence filter**: each sentence gets a probability of containing a word to bleep. Most sentences are cleared at once. 2. **Word-by-word decisions**: the remaining sentences are read word by word. Each word to bleep turns **red** with its category and probability, one after the other. Words to check turn **yellow**. ![Live word-by-word decisions of Smart Profanity](https://mintcdn.com/premieregpt/QduopnJH0A5j0meC/images/smartprofanity-live.png?fit=max\&auto=format\&n=QduopnJH0A5j0meC\&q=85\&s=fcd3c2728227d882c3c4b1039671517d) ## Review After the analysis, every occurrence appears with its timecode, the word in its sentence, its category and its probability. ![Smart Profanity review list with words to bleep and words to check](https://mintcdn.com/premieregpt/QduopnJH0A5j0meC/images/smartprofanity-review.png?fit=max\&auto=format\&n=QduopnJH0A5j0meC\&q=85\&s=a1f6176d8fbcbbdbaa6b0a1032aaff22) * Click a row to check or uncheck it. Click a category pill to switch the whole category. * Click the timecode to move the playhead there. * Click ▶ to listen to the passage with the bleep, before anything is applied. * Under a selected row, **Always bleep this word** and **Never bleep this word** fill your lists and update every occurrence of the word. ### Full transcript Open **Full transcript** to bleep any word yourself, or to unbleep one. Words to bleep are red, words to check are yellow. ![Search bar and full transcript of Smart Profanity](https://mintcdn.com/premieregpt/QduopnJH0A5j0meC/images/smartprofanity-transcript.png?fit=max\&auto=format\&n=QduopnJH0A5j0meC\&q=85\&s=2e82251d263337fc281f6c0735a4cc6a) * **Find a word**: type a word to highlight every match. **Enter** goes to the next match, **Shift+Enter** to the previous one. * **‹ ›** move from match to match. With an empty search, they jump from flagged word to flagged word. * **Bleep all (N)** bleeps every match at once, **Keep all (N)** undoes it. * Click in the transcript, then use **← →** to move the outlined word and **Enter** to bleep or keep it. * Shift-click selects several words of the same sentence. ## Apply When you click **Bleep N words**: 1. Your sequence is **backed up**. 2. The voice is muted on the voice tracks, word by word, with short fades. In Premiere Pro this is done with volume keyframes you can adjust afterwards. 3. Each bleep is created at the exact length of its word and placed on a free audio track, in the **PremiereGPT\_Bleeps** bin. 4. Markers are added if the option is on. > [!NOTE] > > In DaVinci Resolve, the voice is cut on the frame and the bleeped clips are re-placed as audio only on the same track. Their own clip gain and Fairlight effects are reset on those clips. The picture is never touched. To undo, open the backup created before the change. ## Credits | Step | Cost | | - | - | | Transcription | Price of the chosen engine, shown in the engine menu. Always charged. | | Detection | A flat fee, charged only when at least one word is found. | | Applying | Free. | ## Support If you need help, join our [Discord community](https://discord.gg/WyBNCbZhBD) or email us at [admin@premierecopilot.com](mailto:admin@premierecopilot.com). This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.