Who said what?
Transcription APIs that separate the speakers.
Interviews, podcasts and meetings need speaker labels, not one wall of text. Start with documented diarization and compare only estimates that include its known cost.
Developer APIs, not meeting-assistant subscriptions.
Ranked for your volume
Transcription monthly cost by provider
Costs update as you type.
Equal-sized requests; one audio channel.
Melia 1 at Speechmatics
9 priced options · No language requirement · speaker labels included
9 options with a full price · 10 more need a closer look (below)
| Workflow / evidence | ||||
|---|---|---|---|---|
| Melia 1SpeechmaticsLowest estimate | $12.00 | Not benchmarked | $0.12with speaker labels | Recorded filesSources & billing notesPre-recorded pay-as-you-go rate. Multilingual model; speaker diarization included. Price snapshot: 2026-10-17. Sourced preview; confirm the provider quote. Official pricing ↗ · Feature evidence ↗ |
| Universal-2AssemblyAI | $17.00 | Not benchmarked | $0.17with speaker labels | Recorded filesSources & billing notesPay-as-you-go pre-recorded rate per hour of audio submitted. Speaker labels +$0.02/hr. Price snapshot: 2026-10-17. Sourced preview; confirm the provider quote. Official pricing ↗ · Feature evidence ↗ |
| Voxtral Mini Transcribe 2Mistral AI | $18.00 | Not benchmarked | $0.18with speaker labels | Recorded filesSources & billing notesBase transcription estimate; account billing rules can affect the invoice. Price snapshot: 2026-10-31. Tracked dataset. Official pricing ↗ · Feature evidence ↗ |
| Scribe v2ElevenLabs | $22.00 | 5.8%English average · lower is better | $0.22with speaker labels | Recorded filesSources & billing notesListed API transcription usage rate; optional keyterm prompting, transcript editing and plan fees are excluded. Price snapshot: 2026-10-31. Tracked dataset. Official pricing ↗ · Feature evidence ↗Accuracy: Open ASR Leaderboard result for |
| Universal-3.5 ProAssemblyAI | $23.00 | Not benchmarked | $0.23with speaker labels | Recorded filesSources & billing notesPay-as-you-go pre-recorded rate per hour of audio submitted. Speaker labels +$0.02/hr. Price snapshot: 2026-10-17. Sourced preview; confirm the provider quote. Official pricing ↗ · Feature evidence ↗ |
| StandardSpeechmatics | $23.00 | Not benchmarked | $0.23with speaker labels | Recorded filesSources & billing notesPre-recorded pay-as-you-go rate; speaker diarization included. Price snapshot: 2026-10-17. Sourced preview; confirm the provider quote. Official pricing ↗ · Feature evidence ↗ |
| GPT-4o Transcribe DiarizeOpenAI | $36.00 | Not benchmarked | $0.36with speaker labels | Recorded filesSources & billing notesSpeaker labels included. Billed in tokens; OpenAI's own estimate is $0.006 per minute. Price snapshot: 2026-10-17. Sourced preview; confirm the provider quote. Official pricing ↗ · Feature evidence ↗ |
| Amazon Transcribe (batch)Amazon Transcribe | $36.00 | Not benchmarked | $0.36with speaker labels | Recorded filesSources & billing notesBatch transcription, billed per second with no minimum. Speaker diarization, custom vocabulary and language identification are included. Price snapshot: 2026-10-31. Tracked dataset. Official pricing ↗ · Feature evidence ↗ |
| EnhancedSpeechmatics | $38.00 | 6.9%English average · lower is better | $0.38with speaker labels | Recorded filesSources & billing notesHighest-accuracy pre-recorded rate; speaker diarization included. Price snapshot: 2026-10-17. Sourced preview; confirm the provider quote. Official pricing ↗ · Feature evidence ↗Accuracy: Open ASR Leaderboard result for |
Price not fully listed 2
These match your requirements, but the base price or the speaker-label charge is not published in a form we can calculate.
Might fit, not documented 8
The provider does not document at least one of your requirements. That does not mean it is unsupported: test it or ask the provider.
Accuracy is the word error rate (WER) the open Open ASR Leaderboard measured for the exact same model (English average, or FLEURS for the selected language); it is not a test of your audio, and rows without an exact match say so. Costs are API usage estimates, not all-in account bills. Includes documented request minimums and known speaker-label charges when required. Excludes taxes, plan fees, credits, retries, storage and other extras. Incomplete language evidence is kept out of a confirmed shortlist. Test representative audio before integrating.
Speaker labels are not speaker identities
Choose for the conversation you actually have
Diarization labels segments as Speaker 1, Speaker 2 and so on. It does not identify a person by name. Overlapping voices, background noise and short interruptions are the cases to test.
| Endpoint | How it enters this comparison | Evidence |
|---|---|---|
| Mistral Transcribe 2 | Speaker labels in the listed transcription estimate. Recorded files up to 3 hours; do not transfer this claim to the realtime endpoint. | Official announcement ↗ |
| ElevenLabs Scribe v2 | Speaker labels supported. Keyterm prompting, editing and account plan fees are separate from the base API estimate. | Official model details ↗ |
| Deepgram Nova-3 | Feature documented; its add-on price is not confirmed here, so it stays in the price-check group. | Diarization docs ↗ |
| Soniox Async v5 | Speaker separation documented. Token billing is not converted into a made-up fixed hourly rate. | Pricing and included features ↗ |
Before you integrate
Try a two-person interview, a group conversation and a section with interruptions. Check whether labels stay consistent across the file, whether overlapping speech survives and how much cleanup is needed. If you need live speaker labels, switch the workflow to Live; recorded-file evidence is not a promise of realtime support.