POST
/
v1
/
speak
curl -X POST "https://api.bland.ai/v1/speak" \
  -H "authorization: YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "voice": "Maeve",
    "text": "Hello, this is a test of the text-to-speech system.",
    "output_format": "pcm_44100"
  }'
{
  "status": "error",
  "errors": [
    {
      "error": "Invalid Input",
      "message": "Missing text in request body - must be a string"
    }
  ]
}

This endpoint only supports Bland’s Beige Voices.

​
Pricing

Text-to-speech pricing varies by plan:

  • Start Plan: $0.02 per 100 characters (Default)
  • Build Plan: $0.02 per 150 characters
  • Scale Plan: $0.02 per 200 characters
  • Enterprise Plan: $0.02 per 400 characters

​
Headers

​
authorization
string
required

Your API key for authentication.

​
Body Parameters

​
voice
string
required

The ID or name of the Beige Voice to use for speech synthesis.

​
text
string
required

The text content to be converted to speech.

​
output_format
string
default:"pcm_44100"

The audio output format for the generated speech.

Available formats:

  • pcm_8000 - PCM 8kHz, 16-bit, mono
  • pcm_16000 - PCM 16kHz, 16-bit, mono
  • pcm_24000 - PCM 24kHz, 16-bit, mono
  • pcm_44100 - PCM 44.1kHz, 16-bit, mono (default)
  • ulaw_8000 - μ-law 8kHz, 8-bit, mono
​
consistency
float

Controls voice consistency and stability. Value between 0 and 1.

  • 0 - More varied, expressive speech
  • 1 - More consistent, stable speech
​
expressiveness
float

Controls voice expressiveness and emotion. Value between 0 and 1.

  • 0 - More monotone, neutral speech
  • 1 - More expressive, emotional speech
​
stream
boolean
default:"false"

Enable streaming audio response for real-time playback.

When true, audio is streamed as it’s generated. When false, the complete audio file is returned after generation.

​
Response

​
Non-Streaming Response

​
audio
audio/x-wav

Complete WAV audio file with the specified output format containing the synthesized speech.

​
Streaming Response

​
audio_stream
audio/x-wav

Chunked WAV audio stream. The response starts with a WAV header followed by audio data chunks as they’re generated.

​
Response Headers

​
x-latency
string

Time in milliseconds from request to first audio chunk/complete response.

​
x-cost
string

Cost in USD for the text-to-speech conversion

​
Content-Type
string

Always audio/x-wav for successful responses.

​
Content-Length
string

Size of the complete audio file in bytes (non-streaming only).

​
Transfer-Encoding
string

Set to chunked for streaming responses.

curl -X POST "https://api.bland.ai/v1/speak" \
  -H "authorization: YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "voice": "Maeve",
    "text": "Hello, this is a test of the text-to-speech system.",
    "output_format": "pcm_44100"
  }'
curl -X POST "https://api.bland.ai/v1/speak" \
  -H "authorization: YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "voice": "Maeve",
    "text": "This is a streaming example that will return audio chunks as they are generated.",
    "output_format": "pcm_24000",
    "stream": true
  }' \
  --output audio_stream.wav
curl -X POST "https://api.bland.ai/v1/speak" \
  -H "authorization: YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "voice": "Maeve",
    "text": "This example uses custom voice parameters for more expressive speech.",
    "output_format": "pcm_16000",
    "consistency": 0.3,
    "expressiveness": 0.8
  }'
{
  "status": "error",
  "errors": [
    {
      "error": "Invalid Input",
      "message": "Missing text in request body - must be a string"
    }
  ]
}