Text to speech API pricing: what an hour of audio really costs
Text to speech APIs are priced per character, so the bill grows with every line your app says. Comparing them is awkward: one service counts characters, another bytes, a third sells monthly credits. This guide brings them to one unit, an hour of audio, using the list prices on the official pages as of 25 October 2026.
The SuperClonica speech API is priced differently: by time, with a flat monthly plan. Below you will see what that means in hours and where the per-character services are still the better buy.
One unit: an hour of audio
A minute of speech is about 1,000 characters of text, the rule ElevenLabs itself uses on its pricing page. An hour is then about 60,000 characters, and 216 hours about 13 million. The prices below are pay-as-you-go list prices per million characters, without free tiers or volume discounts.
- Google Cloud Standard voices: $4, an hour of audio ≈ $0.24.
- OpenAI tts-1, Azure Neural: $15, an hour ≈ $0.90.
- Amazon Polly Neural, Google Neural2: $16, an hour ≈ $0.96.
- OpenAI tts-1-hd, Google Chirp 3: HD, Deepgram Aura-2, Amazon Polly Generative: $30, an hour ≈ $1.80.
- ElevenLabs Flash / Turbo: $40, an hour ≈ $2.40; Multilingual v2 / v3: $80, an hour ≈ $4.80.
- Fish Audio S2.1 Pro: $15 per million UTF-8 bytes, so a Cyrillic text costs twice as much as an English one and a Japanese or Korean one three times.
The SuperClonica plans in hours
A SuperClonica plan gives minutes of audio in every 10-minute window, for a month bought from your wallet on the site. Used without a break, that is the most a plan can voice in a month:
- Free: 1 minute of audio a day, no card needed.
- Starter, $5 a month: 3 minutes every 10 minutes, up to 216 hours a month.
- Author, $10: 8 minutes every 10 minutes, up to 576 hours.
- Pro, $20: 25 minutes every 10 minutes, up to 1,800 hours.
- Studio, $50: 80 minutes every 10 minutes, up to 5,760 hours.
The same 216 hours elsewhere
At 1,000 characters a minute, the 216 hours of the $5 plan come to about 13 million characters. At list prices that is about $52 with Google Standard voices, $194 with OpenAI tts-1 or Azure Neural, $207 with Amazon Polly Neural, $389 with OpenAI tts-1-hd, Google Chirp 3: HD or Deepgram Aura-2, about $488 on the Cartesia Scale plan, $518 with ElevenLabs Flash and $1,037 with ElevenLabs Multilingual.
The home page of SuperClonica keeps this comparison as a table with every service side by side.
When per-character pricing is the better deal
A flat plan pays off when you voice a lot. If your app says a few sentences a day, a per-character service with a free tier may cost you nothing at all. The big clouds also offer more languages, including Chinese, which Supertonic 3 does not speak yet, SSML, word timings, regional hosting and contracts with uptime guarantees. Compare the sound too: voices differ more than prices, and SuperClonica lets you hear any catalog voice free on its page before you pay.