← All postsDeveloper testing text to speech API output with headphones and waveform on screen

Text to Speech API: Weighted Calls and Trial Limits Explained

Use FreeDailyPro speech synthesize and voices endpoints with callWeight 25, trial limits, and MP3 or WAV output.

text to speech APIspeech synthesizeweighted callsMP3 WAVAPI trialFreeDailyProvoicesaudio toolscallWeightdeveloper
4 min readAudio Tools

Jonah Park ships release notes as short audio for a product team in Seattle. The browser Text to Speech tool is perfect for one-off downloads. The speech API is what he needs when a build pipeline must speak a string and store an MP3 without a person clicking Export. The limits are different from a normal JSON call, and that is the whole point of this post.

Two endpoints

From the live OpenAPI surface:

  • GET /v1/speech/voices — lists voices, formats, limits for your key, and the synthesize weight. Counts as one call. Requires an API key.
  • POST /v1/speech/synthesize — returns the audio file (MP3 by default, or WAV). Requires an API key; there is no key-less synthesize path.

Voices today: amy, lessac, and ryan (American English) and alba (British). Speed range is 0.5 to 2. Headers on a successful synthesize include X-Audio-Seconds, X-Voice, X-Characters, and X-Call-Weight.

Weighted calls in plain numbers

A normal utility call (word count, hash, validate) counts 1 against the monthly allowance. A synthesize call counts callWeight: 25. The voices response documents that number so your client does not guess.

Plans on developers share one pool across REST and MCP:

PlanPriceCalls per month
Trial$020,000
Premium$24 / month200,000
Ultra$69 / month750,000
Mega$129 / month2,000,000 + priority support

Rough math on a paid plan: Premium’s 200,000 calls divided by 25 is 8,000 synthesize calls a month if you spent the whole pool on speech alone. On a trial key the arithmetic is different, because speech has its own cap (next section). In practice Jonah mixes validate and text calls, so he budgets speech separately.

Trial key extras (important)

On a trial key, synthesize is tighter than the monthly pool story:

  • 20 synthesize calls in total (not 20 per month) — field trialSynthesizeCalls on the voices response.
  • Up to 500 characters per synthesize request.
  • Paid plans raise the per-request text cap to 2,000 characters.

Listing voices still costs one ordinary call and is the right first request after you create a key.

Worked example: release-note MP3 in CI

  • Engineer: Jonah Park
  • Text: “Version 2.4 is live. Password reset e-mails now expire in thirty minutes.”
  • Voice: amy, format mp3
  • Expectation: file on disk, headers show character count and call weight 25
  • Browser twin: he still uses Text to Speech when a marketer wants a one-off without keys
curl -X POST https://api.freedailypro.com/v1/speech/synthesize \
  -H "X-API-Key: YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"text":"Version 2.4 is live. Password reset e-mails now expire in thirty minutes.","voice":"amy","format":"mp3"}' \
  --output release-2-4.mp3

Steps

  1. Create a key on developers.
  2. Send a GET to /v1/speech/voices and read maxCharacters, callWeight, and trialSynthesizeCalls if present.
  3. Keep scripts under the character cap for your key.
  4. POST /v1/speech/synthesize with text, voice, and optional format (mp3 or wav).
  5. Save the binary body; read X-Call-Weight to confirm 25.
  6. For longer scripts, split under the character cap and join files with Audio Merger or your own tool chain. Audio Converter helps when a publisher demands a different container.
  7. Browse audio tools for human-side editing after the API returns a file.

Browser tool vs API

NeedUse
One download, no keyBrowser Text to Speech (server voices, MP3/WAV, up to about 3,000 characters per file on the tool page)
CI / backend / agent/v1/speech/synthesize with key and weighted quota
Device voices only, no fileText Read Aloud on the site (different product behavior; no saved file)

Do not assume browser character limits and API trial character limits are the same number.

Common mistakes

  • Surprising finance when 40 narrations consumed 1,000 call units (40 × 25).
  • Burning the 20 trial synthesizes on throwaway tests with 500-character lorem.
  • Omitting the API key and expecting the public “try it now” word-count playground rules to apply—synthesize needs a key.
  • Sending speed outside 0.5–2 or an unknown voice name.
  • Forgetting that MCP video/audio file tools use smaller size limits than REST; speech synthesize returns audio bytes over REST in the example above.

Limits and good to know

  • Synthesize is processed in memory; the OpenAPI text states nothing is stored.
  • Shared monthly pool with every other endpoint and MCP tool.
  • FreeDailyPro offers 200+ professional tools in the browser for non-API work.
  • Overview context: API and MCP server-side tools.

Planning a podcast-style batch

Jonah scripts a nightly job that speaks at most five short strings. He checks remaining trial synthesizes during beta by tracking a counter in his own database—the platform trial synthesize cap is total, not monthly. After upgrade to a paid plan, he still multiplies job size by 25 when comparing to the monthly call ceiling.

FAQ

Why does one MP3 cost 25 calls? Speech synthesis is heavier than a checksum or word count. The API documents callWeight: 25 so billing is predictable.

How many trial synthesizes do I get? 20 total on a trial key, with up to 500 characters each, per the voices response fields.

Can I use the same key for MCP? Yes. REST and MCP share the key and the monthly call pool.

Closing next step

Weighted calls only surprise teams that never read the voices payload. Open developers, send a GET to /v1/speech/voices, then synthesize one sentence to amy and confirm X-Call-Weight is 25.

If you want a sample split-and-merge script for long narrations next, write to https://www.freedailypro.com/contact.

Open Text to SpeechOpen Audio MergerOpen Audio Converter