Start a podcast in your browser: recording, noise removal, editing, and WAV/MP3 export

To start a podcast, record a short spoken take, reduce noise if needed, edit the conversation, and export a WAV or MP3. LA Studio's individual-file noise removal is a free browser processing feature. Batch denoise requires Pro, and cloud saving has login and plan conditions. Those boundaries matter when starting without a production budget: this workflow does not make every recording, storage, and distribution service unconditionally free.

Prepare a small episode you can finish and check

Use your existing microphone, headphones, and browser before buying equipment. Divide a short script into an introduction, main discussion, and ending so retakes are easy to identify. Begin with one speaker and clean the speech before adding music.

A useful first result has understandable words, intact edits, and a downloaded file that plays elsewhere. Processing speed and improvement percentages have not been measured for this guide. Judge the actual recording and device instead of planning around a promised completion time.

Record in the browser and preserve the original

1. Open the DAW editor. Save earlier work before creating a project through Projects in the menu.
2. Create an audio track and enable a recording target. Select the microphone or interface in the input controls and grant browser microphone permission.
3. Press Record and read a short passage after the countdown. Stop, then play the recorded region.
4. If the voice distorts, reduce the input and record again. Address microphone wind and room reflections before recording where possible.
5. Export a usable take as WAV and retain it as the original. Give later processed versions distinguishable names.

If Record is unavailable, check both the target track and active input. A read-only shared project cannot record. For a missing microphone, check permissions and device connection. Listen to the actual region rather than relying on live monitoring. See the home recording instructions for setup diagnosis.

Send a copy of the take to noise removal

Select your recording on Noise Removal to continue into its editor panel. The input is an audio file, but a file picker accepting an extension does not guarantee that the browser can decode it. If loading fails, isolate the problem with a short WAV that plays correctly on your device.

1. Wait for the model status to reach Ready. Model loading and audio decoding are separate steps.
2. Press Denoise. Use Retry for model loading errors; preserve the original before retrying a stalled processing job.
3. After completion, use Continue editing in Editor. The standalone uploaded-file path inserts processed audio on a new track; identify its name.
4. If the original is also present, mute one version and alternate playback over the same passage.

The denoising path converts input to 48 kHz mono for speech enhancement. It does not preserve an interview's separate left/right speakers or a stereo music bed as stereo. Processing speaker files individually makes checking easier. Model downloads need a connection; this individual denoising inference runs in the browser. Saving the resulting project to the cloud is a separate upload.

Listen for preserved words, not only quieter gaps

Compare opening consonants, quiet endings, and words after breaths. A quieter background is not enough if speech becomes metallic or incomplete. There is no strength slider in this panel to follow a generic tutorial against. If the result loses necessary sound, return to the original, choose another take, or record again.

Clipping and strong reverberation are not guaranteed repairs. Repeated denoising cannot restore information absent from the source. Compare at similar listening levels and prioritize intelligibility and natural speech. The noise removal guide provides further comparison checks.

Use Batch denoise for multiple takes with Pro

Batch denoise appears below the individual controls. With Pro, choose Select files… and run Denoise → ZIP. The free plan shows an upgrade prompt here. The individual-file feature's free conditions do not apply to batch processing.

Files are processed sequentially. Successful outputs enter the ZIP with names ending in _denoised.wav. Failed inputs produce _FAILED.txt files containing error details. Completion reports successful and selected file counts. A downloaded ZIP alone does not establish that every take succeeded: inspect both the counts and archive contents.

Give inputs distinct base names before processing. A WAV and MP3 with the same name before the extension can produce the same output name inside the archive. Test failed items individually for decoding and processing rather than repeatedly running every successful take. Batch outputs are also mono, so check positioning separately when assembling the episode.

Edit the spoken sequence

Select the processed audio you intend to use and open its region in Audio Editor. To remove unwanted opening and ending material, select the portion to keep and use Trim. To remove a retake, select the unwanted passage and use Delete, then listen through the sentence around the edit. Trim removes material outside the selection; it is not the same operation as deleting the selection.

For an abrupt boundary, apply a fade to a selected interval and check that it does not obscure the next word. Read the track name before editing so that you know which version is changing. Use Undo when you notice an editing mistake and retain the original recording.

If adding music, put it on another track and lower it until quiet phrases remain understandable. Use your own material or material whose terms you have checked. Balance tracks before using extreme brightness changes to compensate for a quiet voice. For several speakers, audition a short passage from each person in sequence.

Export and verify WAV and MP3

Leave only the intended tracks audible and select Full mix in Export. WAV defaults to 44.1 kHz and 24-bit. The free plan supports WAV at 44.1 kHz with 16/24-bit settings; 48/96 kHz and 32-bit float have Pro conditions. Denoising's internal 48 kHz rate and the final file's export rate are separate settings.

For a distribution MP3, switch the format to MP3. Bitrates are 128, 192, 256, and 320 kbps, with 192 kbps as the initial choice. For MP3, the processing path caps a selected 96 kHz rate at 48 kHz. Retain WAV as your editing copy and produce the needed MP3 without repeatedly editing compressed exports.

Normalize does not guarantee a platform's required loudness. Confirm the setting, export, and listen in another player to speech, edit boundaries, music, and the closing words. Resolve silence or source-loading errors by checking and reloading the affected audio before exporting again. Saving an editable project is separate from downloading the episode audio.

Hand the checked file to your publishing service

Exporting does not publish the show. For a Spotify-hosted show, follow Spotify for Creators' official audio publishing instructions for accepted files and account verification. That upload is separate from browser denoising. Play the actual delivery file before submitting it and confirm that it is the edited episode rather than an earlier take.