Riverside Pro $24 with 15 hours of tracks vs Sume audio detach

Riverside separates tracks at the recording; Sume separates the audio from a video after the fact. Riverside's pricing page (read 2026-10-03) lists Pro at $24 a month on annual billing with 15 hours of separate track downloads, while Sume's audio detach turns one video file into a WAV or MP3 for $0.01 a job. Sume cannot split a mixed recording into one track per speaker; it gives you the track the file already has.

The question has a trend behind it. YouTube's Shorts editor is reported to separate projects into distinct audio and video tracks, per Heyorca's YouTube news roundup (read 2026-10-03; the item sits under October 2025), and that is exactly the kind of structure an audio-first workflow wants. How you get that structure depends on whether you control the recording.

What Riverside lists

On annual billing, Pro is $24 a month with 4K video, unlimited text-based editing and 15 hours of separate track downloads. Grow is $34 with 20 hours of separate tracks. Webinar is $79 with 25 hours. API access is listed on the Business plan, which is custom priced. The monthly-billing prices were not confirmed in the text read.

What Sume's audio detach does

Audio detach takes one video on media.sume.com and returns one audio artifact. The default is sample-exact WAV (pcm_s16le); format: "mp3" gives 128 kbps. channels can be source or mono, sample_rate can be 16000, 44100 or 48000, and range takes a start and optional end in seconds. The caps are a source up to 1800 seconds and an output up to 900 seconds, so a longer track needs a range per 900 seconds. The price is $0.01 per job.

There is no stem or speaker separation in the route; a stereo file with two voices mixed gives you the same mix. If each person was recorded to their own file, run detach on each file. Two 30-minute files need four jobs, two per file, because of the 900-second output cap: $0.04.

import math

def detach_jobs(minutes_per_file, files):
    """Audio detach outputs at most 900 s per job, on sources up to 1800 s."""
    if minutes_per_file * 60 > 1800:
        raise ValueError("source over 1800 s: cut it first")
    return files * math.ceil(minutes_per_file * 60 / 900)

for minutes, files in ((30, 2), (30, 4), (20, 3)):
    jobs = detach_jobs(minutes, files)
    print(f"{files} files x {minutes} min -> {jobs} detach jobs = ${jobs * 0.01:.2f}")

Worked examples

A two-person interview recorded as two 30-minute files costs $0.04 to detach on Sume. Riverside's 15 hours of separate track downloads at $29 cover about 15 such recordings a month if each counts per hour of track, though the page does not say how the hours are counted, so treat that as an assumption. For one recording a month, $0.04 beats $24; for a weekly show you also need the recording tool itself, and Sume does not record.

Now take a finished video from a phone: a 12-minute Reel master with music and voice mixed. Detach gives you the mix. You can use it for speech-to-text (set sample_rate 16000 and channels: "mono"), which video inspect transcribes at $0.01 per audio minute, or as the spine of a Timeline render. It cannot pull the voice out from under the music.

When you want the track structure YouTube is reported to be adding to Shorts, you can rebuild it on the Sume side. Detach the audio once, split it into ranges with timeline audio for $0.01 per job, and render the video slots with the matching source_in values. That gives you separate audio and picture decisions, though the output is one MP4 again.

What Sume does and does not do here

• Sume does not record calls or sessions. Riverside's value is the studio; Sume starts after you have a file.
• Sume does not separate speakers or music from voice. Detach returns the track as it exists in the file.
• Sume does give you precise ranges: range.start and range.end in seconds, with a stable refusal code if the range is empty or past the source.
• Sume checks for audio first: a video with no audio track fails detach_source_has_no_audio, so check probe.has_audio with video inspect (frames: false is enough).

What to do

If you record interviews and want each guest on their own track, record in a tool that does that; Riverside Pro at $24 on annual billing is a straight answer. If you already have files and need a clean WAV for transcription or a spine for a render, use detach at a cent a job and keep each source under 30 minutes.