Bulk narration
Kokoro
American and British English, Spanish, French, Italian, Brazilian Portuguese, Hindi. Speed from 0.5 to 2. Up to 10,000 characters per request.
Text-to-Speech API at $0.77 per Million Characters
Kokoro for bulk narration in 7 language codes, Chatterbox for 23 languages, Qwen3 TTS for cloned and designed voices. All three run on one API key and prepaid credit, and /price gives the cost before each job.
$5 in free credits to start. No credit card required.
- per 1M characters on Kokoro
- $0.77
- Kokoro voices
- 41
- languages on Chatterbox
- 23
TTS API Pricing vs Google, Amazon, Azure, OpenAI and ElevenLabs
List prices from each vendor's pricing page, checked 27 October 2026. Voices differ between models.
| Provider and model | Per 1M characters | Notes |
|---|---|---|
| deAPI · Kokoro | $0.77 | 7 language codes, 41 voices |
| deAPI · Chatterbox | $7.71 | 23 languages |
| deAPI · Qwen3 TTS | $12.86 | 10 languages; preset, cloned and designed voices |
| Google Cloud TTS · Standard / Neural2 / Chirp 3 HD | $4 / $16 / $30 | Price per voice type |
| Amazon Polly · Standard / Neural / Generative | $4 / $16 / $30 | Long-Form voices $100 |
| Azure AI Speech · Neural | $15 | Pay-as-you-go, East US; Neural HD $22 |
| OpenAI · tts-1 | $15 | Deprecated, removed 6 January 2027 |
| fal · Kokoro | $20 | Same open model as deAPI's Kokoro (82M) |
| Deepgram · Aura-2 | $30 | Pay-as-you-go; $27 on the Growth plan |
| OpenAI · tts-1-hd | $30 | Deprecated, removed 6 January 2027 |
| ElevenLabs · Flash / Turbo | $40 | API rate, same on every plan |
| ElevenLabs · Multilingual, v3, v4 | $80 | API rate, same on every plan |
deAPI prices from the /price endpoint on 27 October 2026, Kokoro at speed 1. Azure price from Microsoft's Retail Prices API. OpenAI prices gpt-4o-mini-tts in tokens ($0.60 per 1M text input tokens, $12 per 1M audio output tokens), so it has no per-character rate; it is deprecated with the same removal date.
Cost per Job for Bulk Narration
| Job | Characters | deAPI Kokoro | OpenAI tts-1 | ElevenLabs Flash |
|---|---|---|---|---|
| One maximum-length Kokoro request | 10,000 | $0.0077 | $0.15 | $0.40 |
| Course of 40 lessons, 2,500 characters each | 100,000 | $0.077 | $1.50 | $4.00 |
| A book-length text | 500,000 | $0.39 | $7.50 | $20.00 |
Kokoro, Chatterbox or Qwen3 TTS by Language and Voice
-
-
Many languages
Chatterbox
23 languages including Polish, Arabic, Hindi and Japanese, with one voice in each. Up to 2,000 characters per request.
-
Your own voice
Qwen3 TTS
Nine preset voices, a clone model that copies a 5–15 second recording, and a design model that builds a voice from a description. Up to 5,000 characters per request.
How Billing Works
-
Price before each job
Every task has a /price endpoint, so you can quote a whole batch before it runs.
-
Spend caps per key
A daily, weekly or monthly limit on each key. Over the limit the API answers 402 and the job does not run.
-
Prepaid credit
Jobs draw from credit you top up. A key never spends more than the account balance.
- $5 in free credits
- No credit card required
- Prepaid, no subscription
Price a Narration Job Before You Run It
$5 in credits after email verification, no card.
See also Speech-to-Text API Pricing and AI Voice Generation.
Frequently Asked Questions
Text-to-speech prices, languages and limits
On deAPI, Kokoro costs $0.77 per million characters at normal speed, Chatterbox $7.71 and Qwen3 TTS $12.86. Kokoro's speed setting changes the price. The /price endpoint returns the exact amount.
On list price, yes. ElevenLabs charges $40 per million characters on Flash and Turbo and $80 on multilingual models. deAPI charges $0.77 on Kokoro, $7.71 on Chatterbox and $12.86 on Qwen3 TTS. Listen to the voices before you compare on price.
Kokoro: American and British English, Spanish, French, Italian, Brazilian Portuguese and Hindi, with 41 voices. Chatterbox: 23 languages. Qwen3 TTS: 10 languages.
Up to 10,000 characters on Kokoro, 5,000 on Qwen3 TTS and 2,000 on Chatterbox. Split longer texts into several requests.
Yes. Qwen3 TTS has a voice-clone model that takes a 5–15 second reference recording and a voice-design model that builds a voice from a text description.
MP3 at 24 kHz on all three models.
Yes. deAPI's OpenAI-compatible gateway serves /v1/audio/speech. Change the base URL, the API key and the model and voice names.