Riverside review: what it is good at and where it runs out

Bottom line: Riverside is a strong choice if you record other people remotely: each guest is recorded locally on their own device, you get separate tracks, and a capable text-based editor, hosting and publishing sit in the same account. If you only ever edit your own voice, much of what you pay for goes unused, and its AI voice tools may be more than you want; that is where a narrow editor such as Paraphrase fits.

We make a competing editor, so treat this as an interested party's assessment. We have not used Riverside hands-on for this review. Every claim below comes from Riverside's own pricing page and help centre, read on 17 October 2026.

What Riverside is

Riverside is a remote recording studio that has grown into a content platform. Its editor overview lists timeline editing, transcript editing, captions, AI tools, layouts, music and sound effects, motion graphics and exports, and the pricing page adds podcast hosting, publishing, live streaming, webinars, newsletters and a creator website. There is a free plan, and Pro, the cheapest paid plan, was £35 a month or £24 a month billed yearly as shown to a UK visitor on 17 October 2026.

What Riverside does well

Local recording is the core, and it is the right design. The help centre explains that "High-quality tracks are recorded locally on each participant's computer or device. Slow internet should not affect their quality." For remote interviews this matters more than any editing feature, because you cannot fix a guest's audio that was mangled by their connection.

Separate tracks are first class. Each participant's track can be downloaded raw or aligned, as WAV or MP4, or as compressed MP3, for use in another editor. The help centre says "Your separate, high-quality tracks do not include the watermark."

The text-based editor covers the basics properly. Click a word or drag across a phrase and delete it; cut text stays visible in grey with a strikethrough so you can see what you removed. You can copy and paste text to rearrange the recording, and when two people talk over each other it asks which speaker you meant to cut. Speakers are labelled by their studio display name, unless two people recorded into the same audio input.

Filler word removal is reviewable, not just a switch. The help centre describes a review panel that groups fillers into types: "um", "uh" and "hmm"; contextual fillers like "like", "you know" and "I mean", "flagged only when used as a filler"; breaths; and a "Harsh cuts" group for fillers that might clip a neighbouring word. You accept or discard each one, or remove them all. Flagging risky cuts separately is a thoughtful touch.

There is a tool for whole segments, not just words. Find Fluff, on paid plans, has the Co-Creator suggest "redundant takes, chatter, and technical discussions" as comments you apply or discard. Pause removal has a slider for how aggressive it is.

It is one account for everything. Hosting, publishing to YouTube and Spotify, Magic Clips, show notes and, from the Grow plan up, live streaming are in the same place as the recording. Fewer tools to stitch together.

Transcription covers a lot of languages. The help centre says it transcribes "dozens of languages", from English and Spanish to Welsh and Yoruba.

Where Riverside runs out

None of these are flaws for the people Riverside is built for. They are where its approach stops fitting some readers.

It is priced as a platform. Pro bundles 4K recording, hosting, AI B-roll, eye contact correction and more. If you record alone and only want a clean edit of your own voice, you are paying for a studio you do not use. There is more on this in our Riverside pricing breakdown.

The free plan is a trial of the studio. It lets you try 2 hours of multitrack recording, up to 720p video, and a watermark on exports (on audio, a "Powered by Riverside" track at the start).

Separate track downloads are capped below Business. Recording is unlimited, but downloading per-participant tracks is limited to 15 hours a month on Pro, 20 on Grow and 25 on Webinar.

Some of its AI generates speech. On paid plans, AI Voice turns text into audio "that simulates your speaking voice", voice clones generate audio from a script, and VideoDub replaces up to two words at a time with generated audio and matching lip movement. Riverside documents limits and consent steps for these, and plenty of people want them. If you would rather your edit could only ever contain words you actually said, these features are something to leave alone rather than a reason to choose it.

The AI cleanup is several tools. Filler words, pauses and Find Fluff are separate tools in the editor, each applied and reviewed on its own.

Third-party AI is involved. Riverside's AI page says it uses "third-party AI sub-processors and our own in-house AI", names OpenAI, Eleven Labs and Heygen among them, and says those sub-processors "do not use customer data to train their own models". That is a reasonable policy; it is worth knowing if you care where your recordings go.

Where Paraphrase fits, and where it does not

Paraphrase is narrower. It edits your own recording by its word-timed transcript: click a word to cut it, drag it to move it, and the export is your original audio with pieces removed, as WAV, FLAC, MP3, M4A or Opus, or MP4 for a video take. It never generates speech: no AI voice, no clone, no overdub, by design.

Its AI cleanup is one button. A language model reads the transcript and proposes the whole cut, covering filled pauses, false starts, hedges and repetition, including the same point made twice in different words, and can move words where a sentence reads better. Everything it cuts stays on screen struck through so you can restore it. It runs on Paraphrase's own machines, the transcript is not sent to a third-party model, and it has been tested on English speech. The AI cleanup docs explain it.

Where it does not fill Riverside's role, plainly:

• No remote recording studio and no separate per-speaker tracks.
• No speaker labels, captions, subtitles or translation.
• No noise removal, enhancement, levelling, music or effects.
• No collaboration, comments or share links, and no hosting or publishing.
• 25 European languages for transcription, against Riverside's much longer list.

Premium is £13.99 a month, or about £4.99 a month on the three year cycle, with 120 hours of audio a month. There is a free plan with no card.

Who Riverside is right for

Riverside is right for you if you record guests remotely, want local studio-quality tracks per person, and like the idea of recording, editing, hosting and publishing in one place. It is also the more complete choice if you work in video, need captions, or want AI voice tools.

It is more than you need if you work alone and the edit is the whole job. Our Riverside alternatives post covers the options, and Paraphrase vs Riverside compares the two editors directly.

The quickest way to judge is your own recording. Upload one and press the cleanup button: create a free account at paraphrase.bot, no card needed.