Learn: answers about Speakvora

Short, sourced answers about the Speakvora speech API: how-tos, use cases, pricing and common questions about voices, speech to text and agents. Every page is checked against our product facts before it is published.

What is a text-to-speech API?

What is a text-to-speech API?

What is the difference between text to speech and speech to text?

Text to speech vs speech to text

How is text-to-speech priced?

How text-to-speech pricing works

What is an MCP server for voice?

What is an MCP server for voice?

Add Spoken Feedback to Your App

Turn text into speech with Speakvora's API. Play confirmations and feedback messages to users with library phrases or live generation.

Custom Voice Creation: Record, Upload, or Design

Create custom voices in Speakvora: record a paragraph, upload speech, or design with tuning meters. English supported; French and German beta. Max 3 voices per account.

Create, scope and revoke API keys safely

Manage up to 5 API keys per account with agent scoping, server-side storage, and webhook notifications for key creation and revocation.

Do repeat requests for the same text cost again?

Library hits and repeats are billed by character every time. Custom voice audio stores only after a second request and is deleted after 75 days without use.

WebSocket Streaming: Audio Links and Live Transcription

Speakvora streams speech and transcription over WebSocket with audio returned as URLs per sentence, not raw bytes. Live transcription and agent turns supported.

Download an industry pack after purchase

Get a signed download link for your purchased Speakvora pack using the API, then extract the audio, manifest and phrases.

Browser voice agents with session tokens

Use a 10-minute session token to let a browser talk to a voice agent without exposing your API key.

Text-to-Speech Generation Speed

Speakvora returns library phrases instantly and generates new text in seconds. One test measured 6.5 s for Spanish and 8.7 s for French on 2026-10-08.

Audio and Voice Recording Retention

How Speakvora stores, secures and deletes your custom voice recordings, agent audio, and conversation memory.

API Keys and Scoped Access

Manage up to 5 API keys per account. Scoped keys restrict access to a single agent. Learn authentication, key limits, and webhook events.

How Speakvora Text-to-Speech Billing Works

Speakvora charges per character for text-to-speech: $3 per 1M for catalog voices, $3 per 1M for custom voices. Library hits are billed every time.

SDKs and client libraries for Speakvora

Speakvora offers single-file JavaScript and Python helpers. Call the API directly from any language using HTTP.

Play Speakvora speech in a browser safely

Call /v1/speak from your backend and pass the audio URL to the browser. Agent tokens work in the browser directly but are limited to agents.

Prepare phrases ahead of time with /v1/prepare

Use POST /v1/prepare to queue up to 200 texts for background generation without billing. Ready phrases return instantly; queued texts are ready in a few seconds.

Turn text into speech with one curl request

Make a single POST request to /v1/speak with text and a voice ID to get a URL to audio. Authenticate with x-api-key header.

Transcribe Long Recordings in Pieces

Split audio files over 4.5 MB into smaller chunks for transcription. POST /v1/listen accepts 1 KB to 4.5 MB per request, billed per second rounded up per request.

Give Claude Code or Cursor a voice with Speakvora MCP

Connect Speakvora's text-to-speech API to Claude Code or Cursor using the MCP server. Add voice to your AI assistant in minutes.

Use the Speakvora Python or JavaScript helper

Single-file Python and JavaScript helpers for Speakvora text-to-speech and speech-to-text. No npm or PyPI package; download directly.

Verify Speakvora webhook signatures

Validate webhook authenticity using HMAC-SHA256 signature verification with timestamp and raw body. Reject requests older than 5 minutes.

Speakvora MCP Server for Claude

Use Speakvora's MCP server with Claude to generate speech, list voices, and search phrases via a single setup command.

Write numbers as words for non-English speech

English accepts digits in text-to-speech, but Spanish, French, Portuguese, Hindi, Japanese and Chinese require numbers written as words to generate speech.