> For the full page index, fetch https://echonova.ai/llms.txt # Clean a recording Upload a noisy recording on the Sound Cleaning page, compare the original with the cleaned version, and download the cleaned file. When you finish this page you will have a cleaned copy of a recording: the voice kept, the background dropped, saved as a WAV file on your computer. Clean a file before you transcribe it, before you hand it to an editor, or before you publish audio that was captured in a poor room. ## What you need - A workspace you belong to. Everyone in the workspace can use the page. - Credits. A cleanup is charged by the length of the audio; the rate is on [Sound Cleaning formats, limits and credits](/docs/playground/cleaning-formats-limits-credits). - A recording in MP3, WAV, FLAC or OGG/Opus, up to 1 hour long and up to 150 MB. ## Steps ### Step 1: Open Sound Cleaning In the sidebar under **Playground**, click **Sound Cleaning**. The page opens under the heading **Hear the words, not the room.** and is built around one box with: - A drop area reading **Drop a recording, or click to browse**, with the five format badges MP3, WAV, FLAC, OGG and OPUS and **Up to 1 h and 150 MB per file** under it. - A **Record** button below the word **OR**, with a level meter and a timer. - One row headed **Preset**: four preset buttons with **Studio Master** selected, and **Advanced** at the end. Step 2 covers them. - A status line reading **Waiting for a file** and the **Clean** button. - On the right, a **Cleaned files** list with a **Search cleaned files** field and an **All in Assets** link. Before the first cleanup in the workspace it reads **Cleaned files land here**: _Every cleanup you run shows up here to replay or download._ The page has no sample file, so start with a recording of your own. **All in Assets** opens the Assets page on its **Sound Cleaning** tab; see [Assets and history](/docs/playground/assets-and-history). ### Step 2: Choose a preset, or open Advanced Pick the button that matches the recording before you press **Clean**; the choice applies to the next file you clean. | Preset | Made for | | ------------------- | --------------------------------------------------------------------------------------------------- | | **Studio Master** | Podcasts, videos, calls and general voice. The cleaner's own defaults; selected when the page opens | | **Heavy AC & Wind** | Outdoor, phone audio, loud air conditioning | | **Radio voice** | High-energy voiceovers and audiobooks | | **Deep Voice** | Keeps low baritone frequencies warm | **Advanced** opens a panel under the box headed **Advanced · the parameters behind** followed by the preset name. It holds eight sliders in three groups, **Low cut**, **Loudness** and **Dynamics**, each slider with its value and a hint: **Rumble & hum cutoff**, **Filter smoothness**, **Target volume**, **Maximum peak level**, **Loud spikes trigger**, **Volume smoothing**, **Reaction speed** and **Recovery speed**. A value that differs from the default shows in red. Changes apply as you move a slider and are used for the next file you clean. **Back to the preset** returns every slider to the preset you last picked. Every value and its range is listed on [Sound Cleaning formats, limits and credits](/docs/playground/cleaning-formats-limits-credits#cleaning-settings). **Recommended**: start with a preset and clean once. Open **Advanced** only when the result is not right, and move one slider at a time. **Less effective**: moving several sliders before the first clean; you will not know which one helped. ### Step 3: Drop the file Drag the file onto the drop area, or click the area and choose the file. The status line reads **Checking the file…**. If the file passes, its name replaces the drop area text, the status line reads **Ready to clean** followed by the file name, and the length appears under the box. Nothing is sent yet. Press **Clean** to start the run with the settings you chose. You can also press **Record** and speak. **Stop** ends the take, which is checked like a file and held on the page; press **Clean** to send it. A take stops on its own at 1 hour. If the microphone cannot be used, the status line reads **Microphone access was not available.** Three checks run before anything is sent. A file that fails one is refused: the status line shows the message in red, and nothing is charged: | Check | Message | | --------------------------------------------- | -------------------------------------------------------------------- | | Extension is one of MP3, WAV, FLAC, OGG, OPUS | **That file type is not supported. Use MP3, WAV, FLAC or OGG/Opus.** | | Size is 150 MB or less | **That file is over 150 MB. Trim it or export a smaller version.** | | Length is 1 hour or less | **That clip is longer than 1 hour. Split it and try again.** | If the browser cannot read the file's length, the status line shows **We could not read that file**. Export the recording again as MP3 or WAV and try once more. ### Step 4: Wait for the cleaned file Press **Clean**. The button reads **Working** with a spinner, the preset row and sliders are disabled, and the status line moves through two states: | Status line | What it means | | ---------------------------- | ----------------------------------------------------------------- | | **Uploading**, then a number | The file is on its way. The number is the percentage sent so far. | | **Cleaning**, then the name | The file has arrived and is being cleaned. | If the run stops, the status line shows the reason in red and **Try again** appears beside **Clean**; it sends the same file again. Every message is listed under **Troubleshooting** below. The charge is taken when the file is accepted after upload and is refunded if the run stops. A file you have already cleaned with the same settings cannot be sent again: change a preset or a slider, or choose another file. ### Step 5: Compare the original with the cleaned version When the cleaned file arrives, the status line reads **Cleaned** followed by the file name, and a **Cleaned** card opens below the box. It shows the file name, then the preset name, the length as `m:ss`, the size of the cleaned file and its loudness in LUFS. Inside it: - Two rows, **Original** and **Cleaned**, each drawn as a waveform of that track over time. The row you are listening to is highlighted. **Time zoom** magnifies the view, and **Emphasize quiet detail** enlarges quiet sounds on screen without changing the audio. When the browser cannot read one of the files, the rows read **Waveform unavailable**; playback is not affected. - **Play** and **Pause**, a **Volume** slider, **Back 5 seconds**, **Forward 5 seconds**, and the time as `m:ss/m:ss`. - **Listen to**, with **Original** and **Cleaned**. **Cleaned** is selected when the card opens. Switching keeps the position, and if the audio was playing it goes on playing in the other track. - Click anywhere on either row to play that track from that point. A file you have just cleaned waits for you to press **Play**. In the **Cleaned files** list, a row opens that cleanup in the card, and the play button on a row opens it and starts the cleaned track at once. ### Step 6: Download the cleaned file Click **Download** in the card header. The file is saved as `cleaned_` followed by your original file name without its extension, and `.wav`, so `interview.mp3` is saved as `cleaned_interview.wav`. It is a WAV file. The original is not offered for download; you already have it. Two more buttons in the same header take the cleaned audio on: **Transcribe cleaned audio** opens Speech to Text with the file already chosen, and **Clone the voice** opens voice cloning with it. To clean another file, drop or record it. The **Clean** button stays off until the file or the settings change. ## Get it right The 150 MB limit is checked on the file as you chose it, so an uncompressed stereo recording crosses it long before the 1-hour limit. **Recommended** - A 1-hour recording exported as MP3, or as 16 kHz mono WAV, which stays inside 150 MB. **Less effective** - The same hour as 44.1 kHz stereo WAV, which is over 150 MB and is refused before upload. ## Worked example You have `interview.mp3`, 12 minutes and 24 seconds recorded in a café. 1. Open **Sound Cleaning**, drop the file on the drop area and press **Clean**. 2. The status line reads **Uploading** with the percentage climbing to 100%, then **Cleaning** followed by `interview.mp3`. 3. The status line reads **Cleaned** and the **Cleaned** card opens. It shows the preset name, `12:24` and the size of the cleaned WAV file. Press **Play** and switch between **Original** and **Cleaned** to hear what was removed. 4. Click **Download**. Your downloads folder now holds `cleaned_interview.wav`. 5. The cleanup cost 12,400 credits, for 744 seconds of audio at 1,000 credits a minute. The sidebar counter labelled **Used** shows the new total. ## What happens next - **Cleaned files** lists the newest 8 cleanups made from the portal in this workspace, by anyone in it, newest first. Each shows the file name, its length as `m:ss` and the date and time, such as `15 Sep, 15:42`. Type in **Search cleaned files** to filter the list. Click a row to load it into the card, or its play button to load it and play the cleaned track. - The credit counter in the sidebar, shown as used out of total with the label **Used**, refreshes when the cleaned file arrives. - Cleanups stay listed until you delete them. They also appear on the **Assets** page under **Sound Cleaning**, where **Delete** removes one from the workspace at once. Deletion is final. - A run that stops before it finishes leaves nothing in **Cleaned files**, and its charge is refunded. ## Limits | Limit | Value | Applies to | | --------------- | ------------------------ | ------------------------------------------------------------------------------------------------------------------------------- | | Formats | MP3, WAV, FLAC, OGG/Opus | Files dropped on the drop area, by extension | | File size | 150 MB | Every file, checked before upload and again after it | | Audio length | 1 hour | Every file, checked before upload and again after it | | Files at a time | One | The drop area takes one file per drop | | Cleaning time | 10 minutes | A run that has not finished by then stops with **Sound cleaning is temporarily unavailable. Please try again.** and is refunded | | Output format | WAV | Every cleaned file | | Cleaned files | The newest 8 | The list on this page | Every message and the price in credits are on [Sound Cleaning formats, limits and credits](/docs/playground/cleaning-formats-limits-credits). ## Troubleshooting Every message below appears in the status line under the box, in red. After a message that arrives after upload, **Try again** appears beside **Clean** and the message is also shown as a notice titled **Error**. After a check before upload, nothing was sent, so drop another file instead. ### That file type is not supported. Use MP3, WAV, FLAC or OGG/Opus. The file's extension is not `.mp3`, `.wav`, `.flac`, `.ogg` or `.opus`. Export the recording in one of those formats. Nothing was sent or charged. ### That file is over 150 MB. Trim it or export a smaller version. The file is larger than 150 MB. Export it as MP3, or split it. Nothing was sent or charged. ### That clip is longer than 1 hour. Split it and try again. The recording is longer than 1 hour. Split it into parts of 1 hour or less and clean each part. Nothing was sent or charged. ### We could not read that file The browser could not decode the file, so its length could not be checked. Export the recording again as MP3 or WAV. Nothing was sent or charged. ### Audio exceeds the 1-hour limit. The browser reported no length for the file, so the file was sent and its length was checked by decoding after upload. Split the recording into parts of 1 hour or less. Nothing was charged. ### Unsupported file type: … Please upload an audio file. The file passed the checks in the browser, but the type it carried is not an audio type; the message names that type. Export the recording again as MP3 or WAV and drop it again. Nothing was charged. ### Could not read the audio length. Try another file. The file was sent but the server could not decode it to measure its length, so the run stopped before anything was charged. Export the recording again as MP3 or WAV and drop it again. ### Insufficient subscription balance The workspace balance cannot cover this file. Nothing was charged. Add credits, or clean a shorter file. **Try again** sends the same file and fails the same way until the balance changes. ### Subscription has expired There is no paid plan and no extra credits are left, so basic limits apply with no plan credits. Choose a plan or add credits under **Subscription & billing**, then **Try again**. ### Sound cleaning is temporarily unavailable. Please try again. The run could not be completed, or took longer than 10 minutes. The charge was refunded. Wait a moment and click **Try again**. ### Cleaning stopped before it finished. Try again. The connection dropped before a result arrived; a notice titled **Error** reading **Unable to connect to the server. Please check your internet connection.** appears as well. Reload the page: if the file is listed under **Cleaned files**, the cleanup finished and was charged, so open it there. If it is not listed, nothing was charged; check your connection and drop the file again. ### Play stops at once, or Download opens a page with an error instead of the file The page has been open for a long time and the item's links have gone stale. Reload the page and open the item again. ### The browser played the file in the tab instead of saving it Save it from that tab with your browser's save command, then go back to the portal. ## FAQ ### Is the cleaned file in the same format as my upload? No. Every cleaned file is WAV, and the downloaded file is named `.wav`. ### Does replaying or downloading cost credits? No. Only the cleanup itself is charged, once, when the file is accepted. ### Can other members see my cleanups? Yes. **Cleaned files** lists every cleanup made from the portal in the workspace, whoever ran it. ### Can I clean a file longer than 1 hour? No. Split it into parts of 1 hour or less and clean each one. ### Can I delete a cleanup? Yes, on the **Assets** page: press **Delete** on it under **Sound Cleaning**, then confirm. Deletion is final. ### Can I clean audio from my own code? Yes. The API cleans a file of up to 10 minutes and 150 MB with one call to `POST /v1/sound-cleaning` and lists, reads and deletes jobs under `/v1/sound-cleaning/jobs`. The key needs Sound Cleaning access. Start at the [Sound Cleaning quickstart](/docs/developers/clean-quickstart). ## Related - [Sound Cleaning formats, limits and credits](/docs/playground/cleaning-formats-limits-credits): every limit and message, and what a cleanup costs - [Transcribe a file](/docs/playground/transcribe-a-file): the next step for a cleaned recording - [Assets and history](/docs/playground/assets-and-history): where speech, transcripts and cleanups are kept - [Credits](/docs/developers/credits): what a credit is and how a charge is counted