Text to speech API prices, per minute of audio
Stork's Speech API is not the cheapest text to speech API per minute: the large cloud providers charge less for their basic voices. It is built for a different job. One request takes a whole script of up to 20,000 characters and returns one MP3, you can use your own voice at the same price, there is no plan or minimum, and the same key turns that audio into a talking video. It costs $0.05 per minute of finished audio. The table lists what we could read on each provider's own pages.
What it costs, next to published prices
| API | Price and limits |
|---|---|
| Stork Speech API | $0.05 per minute of finished audio, billed per started 10 seconds. No plan. 20,000 characters and 30 minutes of audio per request. Your own voice at the same price. Free: 3 files a day. |
| Google Cloud Text-to-Speech | $4 per million characters (Standard, WaveNet), $16 (Neural2), $30 (Chirp 3 HD): about $0.004 to $0.027 a minute. 5,000 bytes per request. |
| Amazon Polly | $4 per million characters (Standard), $16 (Neural), $30 (Generative): about $0.004 to $0.027 a minute. 3,000 billed characters per synchronous request. |
| OpenAI (tts-1, tts-1-hd) | $15 and $30 per million characters: about $0.014 and $0.027 a minute. 4,096 characters per request. |
| Azure AI Speech (Neural) | $15 per million characters: about $0.014 a minute. Up to 10 minutes of audio per real-time request. |
| Deepgram Aura-2 | $0.030 per 1,000 characters: about $0.027 a minute. 2,000 characters per request. |
| Cartesia Sonic | Monthly plans with credits, from $5 for 100,000 characters: about $0.045 a minute on that plan, about $0.034 on the largest. |
| ElevenLabs | $0.04 or $0.08 per 1,000 characters depending on the model: about $0.04 to $0.07 a minute. Pay as you go. 5,000 to 40,000 characters per request depending on the model. |
Other companies' prices were read on their own pages on November 15, 2026 and may have changed. Sources are listed below. Stork's prices are current.
Use Stork when
- You turn long scripts into one audio file: a chapter, an episode or a lesson in a single request, with no chunking or stitching.
- You want your own voice without a higher plan: add it once in the dashboard and it costs the same as a built-in voice.
- You do not want a subscription or a monthly minimum. Credits are prepaid and never expire.
- You also need the video: the Talking Photo API takes the same text and voice and returns a lip-synced MP4.
Use something else when
- Price per character at high volume decides it: the basic cloud voices cost a fraction of a cent a minute.
- You need speech as it is generated, for a live voice agent. Stork makes a finished file and takes from seconds to minutes.
- You need a language outside the 13 that Stork's built-in voices cover.
One request
Add "tier": "free" for a free file while you test.
curl -X POST https://api.stork.ai/api/v1/speech \
-H "Authorization: Bearer sk_stork_YOUR_KEY" \
-H "Content-Type: application/json" \
-d '{
"text": "Welcome back to the channel. Today we are looking at three small habits.",
"voice_id": "en-ava",
"webhook_url": "https://example.com/hooks/stork"
}'The response is a job id. When the MP3 is ready, your webhook gets a signed POST, or you poll the id. Idempotency keys, an estimate endpoint and an OpenAPI file are in the API docs.
Questions
- How do you turn a price per character into a price per minute?
- We assume 900 characters for one minute of English speech and multiply. It is our arithmetic, not a figure the providers publish, and real speech varies with the voice and the text. Stork bills by the length of the finished audio, so its price per minute is exact.
- Is Stork the cheapest text to speech API?
- No. On November 15, 2026 the basic voices of the large cloud providers cost well under one cent a minute by our arithmetic, and several neural voices cost about one to three cents. Stork is $0.05 a minute. Choose it for long files in one request, your own voice at no extra price, and no plan.
- How long can one request be?
- Up to 20,000 characters and 30 minutes of audio in one file. Most of the APIs in the table accept between 2,000 and 5,000 characters per request, so a long script has to be split and the pieces joined.
- Can I use my own voice through the API?
- Yes. Add your voice once in the Stork dashboard from about 15 seconds of speech, then pass its id as voice_id. It costs the same as a built-in voice. Creating a voice is done in the dashboard, not through the API.
- Does it stream audio in real time?
- No. You create a file, then poll or receive a signed webhook when it is ready. Short files are ready in under a minute.
Sources
- Google Cloud Text-to-Speech pricing
- Amazon Polly pricing
- Amazon Polly limits
- OpenAI API pricing
- OpenAI text to speech guide
- Azure AI Speech pricing
- Azure AI Speech quotas and limits
- Deepgram pricing
- Cartesia pricing
- ElevenLabs API pricing
More comparisons: HeyGen API alternative · D-ID API alternative · Talking head API · AI music video API prices
