How much will TTS actually cost?
Get a recording quote before you buy credits. Paste your script, choose a hosted voice preset, and see how the per-recording minimum affects your total.
Price your actual script
OpenSpeech charges by characters, with a minimum for each recording. This calculation runs on your browser.
83 characters · 83 UTF-8 bytes · 2,000 characters / 4,000 bytes maximum per recording
$0.01
1 credits total · $0.01 per recording · 1-credit minimum each
Prepaid packs start at $5. This is usage cost, not the minimum checkout amount. Each recording is started separately.
Copy your script, then paste it into the studio to review the price before creating audio.
Why minimums change the total
Quote and create a multi-paragraph voiceover →Ten short recordings can cost more than one long recording with the same total characters. Kokoro charges 2 credits per 1,000 characters, with a 1-credit minimum for each job. Ten 100-character recordings cost $0.10; one 1,000-character recording costs $0.02. Paragraph projects round each paragraph separately.
Playback length varies with the voice, punctuation and speaking speed. Minutes are not OpenSpeech billing units.
Comparing other providers
Provider subscriptions, character billing and compute time cover different services. Check the selected model, included allowance, overages and commercial terms at the source.
Self-hosting also includes idle capacity, storage, deployment and maintenance. There is no universal cheapest option. OpenSpeech rate snapshot: 2026-09-20.1; the studio confirms the current charge.
- One credit is $0.01. Rates apply to each separately generated recording; prepaid credit packs start at $5.
- Leading and trailing whitespace is removed before counting. Both character and UTF-8 byte limits apply to your script.
- The calculator runs locally in your browser. Your script is not sent to analytics or included in the studio link. Copy and paste it into the studio when you are ready.
- The studio confirms the current price before you generate. Each listed model uses its hosted preset; directory samples may demonstrate other voices or features.