A speech API alternative where you can pay once
Per-character API for new text, one-time bundles for lines that repeat, and the arithmetic for choosing.
Written by the Speakvora team. Prices for other companies come from their own pricing pages, last read on October 8, 2026 (the same table as ). They change, so check before you decide.
The problem with paying by the character
Most speech APIs charge for every character you send, so the bill follows your traffic. That is fine when you are small and a surprise when you are not. If you are looking for something else, it is usually because an invoice came in higher than planned, or because the thing needs to keep working without a connection.
We built two ways to pay, because the right answer depends on your text.
If the text changes: the API
Catalog voices are $3 per million characters. Your own cloned voice is $6. There is no monthly minimum and no plan to outgrow. A new account gets 100,000 characters of stored library phrases free, no card.
For comparison, here is what other providers list for the same million characters.
| Provider and model | 1M characters | 10M characters |
|---|---|---|
| Speakvora catalog voices | $3 | $30 |
| Amazon Polly Standard | $4 | $40 |
| Google Cloud Standard | $4 | $40 |
| OpenAI tts-1 | $15 | $150 |
| Deepgram Aura-1 | $15 | $150 |
| Fish Audio | $15 | $150 |
| Inworld Flash | $15 | $150 |
| OpenAI tts-1-hd | $30 | $300 |
| Rime Mist v3 | $30 | $300 |
| ElevenLabs Flash / Turbo | $40 | $400 |
| ElevenLabs v2 and v3 | $80 | $800 |
This is cost only. These models sound different, and we are not saying ours beat theirs. Listen to the free samples and decide.
If the text repeats: pay once
A greeting, a confirmation, a hold message, an error prompt. These do not change, so paying for them every time they play makes little sense. A bundle is a set of checked lines for one industry and language, recorded once. You buy it once per location and play the files as often as you like.
Rough arithmetic, at list price: $214 buys about 2.7 million characters at ElevenLabs v3 rates, or about 54 million at Polly Standard rates. If you never regenerate a line, the API is cheaper than the bundle. We say that on purpose.
What the bundle gives you is different: the lines are already written and checked, they play offline with no network, the bill is fixed, and the $214 to $399 versions include the local agent and its workflows for texts, bookings and reminders.
Which one to pick
- New sentences every time, or your own voice: the API.
- A kiosk, device, IVR or app that says the same things all day, maybe with no signal: a bundle.
- A small business that wants a phone and text front desk it can set up once: the bundle with the agent.
- Both: bundle for the fixed lines, API for the rest.
Read next: pay-once text to speech, , , how the library grows.
Questions
Is there a speech API alternative where you pay once?
Partly. Speakvora sells two things. The API is pay as you go at $3 per million characters with no monthly minimum. And there are one-time voice bundles for the lines you play again and again: audio from $14, or $214 to $399 per location with our local agent, nothing per play afterwards.
Is pay once cheaper than a speech API?
Not always, and we would rather say so. If you generate a line once and keep the file, per-character pricing costs almost nothing, even at ElevenLabs rates. A bundle makes sense when you want the lines already written and checked for your industry, want them to play with no internet, or want a fixed bill that cannot grow.
What does ElevenLabs cost per million characters?
On its pricing page, Flash and Turbo models are listed at $40 per million characters and the v2 and v3 models at $80, read on October 8, 2026. Plans, credits and limits differ, so check the page for what you would actually pay.
Can I use both the API and a bundle?
Yes, most people do. The bundle covers the lines that never change, and the API speaks whatever is new. A free key includes 100,000 characters of stored library phrases and needs no card.
What are the downsides of Speakvora?
The voices are AI-generated, and we do not claim they beat the best voices from the bigger companies. We are smaller, with fewer languages and models. Non-English bundles are still waiting for native-speaker review. And the one-time price does not cover your phone or text provider.