Text to Speech API Pricing: Compare TTS Costs Fish Audio

The Fish Audio text to speech API costs $99 to start: the Studio pack includes API access, 15,000 credits and roughly 15 million characters of generated speech. If you only need browser generation, packs start at $9; the $99 Studio pack is the entry point when your own product must call the API.

That difference matters because many “TTS API pricing” pages quote a raw per-character rate while omitting the cost of custom voices, concurrency, commercial rights and failed iterations.

How is Fish Audio API usage priced?

One credit covers about 1,000 characters. The Studio pack contains 15,000 credits, so its face value is about 15 million generated characters. At $99, that is approximately $6.60 per million characters if the full balance is used for text to speech.

The same balance can be spent on other supported capabilities. Regenerations consume credits, so the practical cost is the number of takes, not just the final character count.

Commercial usage rights start with Creator. API access, the highest concurrency for real-time streams and 15,000 credits arrive together in Studio. Purchased credits remain valid for 12 months.

How much speech does a character budget produce?

Plan roughly 9,000 to 10,000 characters per hour of finished narration. Speaking pace and punctuation change that number, so use your own script as the final benchmark.

A 100-million-character project is approximately 100,000 credits, or roughly 6.7 Studio packs before revisions. At that scale, negotiate around your actual production plan and verify the current pack limits in the workspace before prepaying.

How do competitor API prices compare?

Vendor rates change, so convert every option to the same “per one million characters” unit and check the date on the source.

These prices are not an apples-to-apples ranking. Fish Audio’s rate includes access to its voice library, cloning workflow and expressive script controls under the Studio pack. Competitor terms, concurrency and voice classes differ.

Use the OpenAI pricing page, Murf pricing page and Amazon Polly pricing page as primary references.

What hidden costs should enter the estimate?

The first generation is rarely the final take. Add a regeneration allowance for names, pacing, pronunciation and copy edits. A 20 percent revision buffer is a practical starting point for a new voice, and some projects need more.

Concurrency is the second cost. A batch job can wait in a queue, but a live voice agent has to answer while the user is listening. The Studio pack carries the highest concurrency on this site, yet your end-to-end architecture still determines whether the experience feels responsive.

Commercial rights are the third. A cheaper development rate is irrelevant if the plan does not allow the intended client, advertising or monetized output. The commercial use guide explains the Fish Audio boundary.

How should an engineering team budget?

1. Count the finished characters per month.
2. Add expected regeneration and failed-job overhead.
3. Separate asynchronous narration traffic from latency-sensitive speech.
4. Check the price of the required voice class, not the cheapest voice headline.
5. Confirm API limits and commercial rights.
6. Recalculate after a one-week production sample.

The text to speech API guide covers request design, streaming, retries and key storage. For conversational workloads, the real-time text to speech guide explains where latency accumulates.

When is Fish Audio not the cheapest choice?

Choose Amazon Polly Standard or Neural when a stock voice and a low per-character rate satisfy the application. Choose OpenAI when its preset voices and existing SDK integration reduce engineering work. Choose Murf when its studio workflow or enterprise controls are the deciding feature.

Choose Fish Audio when voice variety, reusable cloning, expressive directions or one-time credit capacity materially reduce production time.

What budget mistakes are most common?

The first mistake is comparing a subscription allowance with a prepaid character balance without checking which capabilities consume each unit. The second is omitting regeneration, which is part of normal audio production rather than an exceptional failure.

The third is treating concurrency as free. Interactive traffic needs spare capacity, cancellation and monitoring even when monthly character volume looks small. Build the estimate around the busiest hour and the approved voice, not around an average month at the cheapest model.

Key takeaways

• Fish Audio API access starts with the $99 Studio pack.
• Studio contains 15,000 credits, roughly 15 million characters.
• The effective text-to-speech rate is about $6.60 per million characters when the full balance is used.
• Plan a revision buffer because every successful regeneration consumes credits.
• Compare voice class, rights and concurrency, not only the lowest character rate.

Frequently asked questions

How much does the Fish Audio API cost?

API access comes with the $99 Studio pack, which contains 15,000 credits. One credit covers about 1,000 characters, so the pack covers roughly 15 million characters.

Is there a cheaper way to use Fish Audio without the API?

Yes. Browser generation starts with the $9 Starter pack and 1,000 credits. The Studio pack is required when your own application needs API access.

How many characters are in an hour of speech?

Plan about 9,000 to 10,000 characters for an hour of narration. A five-minute script usually falls between 4,000 and 5,000 characters, depending on speaking pace.

Are credits consumed when I regenerate audio?

Yes. Each successful generation spends credits, so script review and smaller generation blocks reduce the cost of revisions. Credits remain valid for 12 months.

Which TTS API is cheapest?

The cheapest listed character rate depends on the voice class and usage pattern. Amazon Polly Standard and Neural have lower per-character rates, while Fish Audio, OpenAI and Murf bundle different voice, cloning and production capabilities.