To automatically remove filler words from a podcast, an AI transcribes your audio, flags each filler (um, uh, like, you know, basically, actually) in the transcript, and trims the audio at those exact timestamps so the words disappear without you touching a waveform. DoneCast does this automatically as part of its cleanup: it detects the fillers in your transcript, cuts them out of the audio, and lets you review every change before you publish.
The process starts with a transcript. When your recording is transcribed, every word gets a timestamp, so the software knows exactly where each 'um' or 'uh' lands in the audio. From there, the AI scans the transcript for known filler words and marks their positions.
Once the fillers are flagged, the tool trims the audio at those timestamps instead of asking you to hunt for them on a waveform. The words are cut, the surrounding speech is joined back together, and what's left sounds like you simply didn't say them.
In DoneCast, filler removal is part of the AI cleanup that runs on your recording, so there's no separate step to configure. It detects the common fillers, um, uh, like, you know, basically, and actually, and trims the audio at those points automatically.
The same cleanup pass handles two related messes: false starts, where you repeat a single word like 'I I I think' and it keeps just 'I think,' and excessive silence, which gets trimmed while your natural pauses stay intact. You can review everything before it goes live, so nothing is cut without your say-so.
The goal isn't to strip every pause and breath, that's what makes edited audio sound robotic. DoneCast trims excessive silence but preserves the natural pauses that make speech sound human, and it targets fillers rather than the words around them.
Because you can review the cleanup before publishing, you stay in control. If a 'like' or an 'actually' was doing real work in a sentence, you can catch it in review rather than discovering it after the episode is out.
Filler removal is one piece of a larger AI podcast host. In DoneCast you record, edit by voice, then publish to Spotify and Apple over RSS, all in one place. You can say 'mulligan' to undo a flub as you record, or say 'Intern, [your question]' to drop in an AI-spoken answer, and the cleanup handles the fillers behind the scenes.
Plans start at $19/mo (Starter), $49/mo (Creator), and $119/mo (Studio), with a 14-day free trial and no card required. Other tools like Descript, Cleanvoice, Riverside, and Auphonic also offer filler removal; DoneCast bundles it into a record-to-publish workflow so you don't stitch separate tools together.
DoneCast detects and removes um, uh, like, you know, basically, and actually by finding them in your transcript and trimming the audio at those timestamps.
No. DoneCast targets the fillers themselves and preserves your natural pauses. It only trims excessive silence, so the cut speech joins back together and sounds like you never said the filler.
Yes. Filler removal is part of DoneCast's AI cleanup, and you can review every change before publishing, so nothing is cut without your approval.
Yes. The same cleanup pass removes single-word false starts, turning 'I I I think' into 'I think,' and trims excessive silence while keeping your natural pauses.
Plans are Starter at $19/mo, Creator at $49/mo, and Studio at $119/mo. There's a 14-day free trial and no card required to start.
Record, edit by voice, and publish your podcast in one place. 14-day free trial, no credit card.
Start Free Trial