Muse Voice Transcribe at $0.18 an hour: 1,000 short clips on Sume

Meta lists Muse Voice Transcribe at $0.18 per hour, which is $3.00 per 1,000 minutes. Sume speech-to-text lists $0.01 per audio minute, which is $0.60 per hour and $10.00 per 1,000 minutes. For 1,000 clips of 20 seconds (333.3 minutes), that is $1.00 on Meta's rate and $3.33 on Sume's. If those clips get burned-in captions on Sume, the $0.20 caption job is the larger line, not the transcript.

I read Meta's Muse Voice Transcribe page, Sume's video inspect docs (the STT rate) and Sume's video captions docs on 2026-10-03. This compares list prices only. It says nothing about accuracy, and I did not run Muse.

What are the two list prices?

Meta's page shows one model, muse-voice-transcribe-1.0, at $3.00 per 1,000 minutes and $0.18 per hour. Sume's docs give $0.01 per audio minute and tell you to confirm the live figure in GET /v1/catalog.

Why does clip length change the bill?

Sume's STT request takes an optional duration_seconds used to reserve usage, and the docs say that omitting it reserves 1 minute. For short clips, send the real length so the reservation matches the clip. Meta's page does not say how it rounds short audio, so check that before you multiply by 1,000.

The code below reproduces the table. Change the two numbers at the top for your own clips.

clips, seconds = 1000, 20
minutes = clips * seconds / 60
meta, sume = 0.18 / 60, 0.01
print(f"{minutes:.1f} min: Meta ${minutes * meta:.2f}, Sume STT ${minutes * sume:.2f}")
print(f"1 hour: Meta ${60 * meta:.2f}, Sume STT ${60 * sume:.2f}")
print(f"Captioned on Sume: ${clips * 0.20:.2f} for {clips} caption jobs")

When is the transcript not the main cost?

A standalone Sume caption job lists $0.20 for a video up to 60 seconds, and the docs show no separate transcription line inside it. So 1,000 captioned clips are $200.00 on Sume, whatever the STT rate is. A transcript-only workflow, such as search or notes, is where the per-minute gap matters.

If you already have word timings, you can pass them in the words field of the caption request and Sume skips speech-to-text entirely, per the video captions docs.

What should you decide before comparing?

Settle these before you let a price gap pick the vendor, because the cheaper line can be the wrong comparison.

• Do you need text only, or a rendered captioned video? Those are different products.
• Measure accuracy on your own clips before price decides; the method is in how to measure word error rate.
• Check each vendor's rounding for clips under a minute.
• Re-read both prices on the day you buy; Sume's docs point to GET /v1/catalog for the live figure.