Skip to main content
POST
https://ai.kaiho.cc
/
v1
/
audio
/
speech
TTS Text-to-Speech
curl --request POST \
  --url https://ai.kaiho.cc/v1/audio/speech \
  --header 'Content-Type: application/json' \
  --data '
{
  "model": "<string>",
  "input": "<string>",
  "voice": "<string>",
  "speed": 123
}
'

​
Overview

The TTS (Text-to-Speech) API uses advanced speech synthesis technology to convert text into natural, fluent human voices, supporting multiple languages and voice options.

​
Supported Voices

Alloy

Neutral, balanced voice

Echo

Male, steady voice

Fable

Male, expressive voice

Onyx

Male, deep and powerful

Nova

Female, friendly and warm

Shimmer

Female, soft and sweet

​
Request Parameters

​
model
string
required
TTS model:
  • tts-1 - Standard quality, fast speed
  • tts-1-hd - High definition quality, more natural
​
input
string
required
Text to convert to speech, up to 4096 characters.
​
voice
string
required
Voice to use: alloy, echo, fable, onyx, nova, shimmer
​
speed
number
default:1
Speech speed, range 0.25 - 4.0.
  • 1.0 = Normal speed
  • 0.5 = Half speed
  • 2.0 = Double speed

​
Request Example

import openai
from pathlib import Path

client = openai.OpenAI(
    api_key="YOUR_API_KEY",
    base_url="https://ai.kaiho.cc/v1"
)

# Generate speech
response = client.audio.speech.create(
    model="tts-1-hd",
    voice="nova",
    input="Welcome to Kaihoxz text-to-speech service. We provide the most natural AI voice synthesis technology.",
    speed=1.0
)

# Save to file
speech_file_path = Path("speech.mp3")
response.stream_to_file(speech_file_path)

print(f"Speech saved to: {speech_file_path}")

​
Voice Characteristics

VoiceGenderCharacteristicsUse Cases
AlloyNeutralClear, professionalNews, education, customer service
EchoMaleSteady, authoritativeBusiness, broadcasting
FableMaleLively, expressiveStorytelling, advertising
OnyxMaleDeep, powerfulDocumentaries, serious content
NovaFemaleFriendly, warmAssistants, guidance
ShimmerFemaleSoft, sweetChildren’s content, casual scenarios
Usage Guidelines: Please ensure that generated speech content complies with local laws and regulations. It must not be used for fraud, impersonation, or other illegal purposes.
Performance Optimization: For real-time applications, use the tts-1 model with opus format to achieve the lowest latency.