Fish AI

The FishAudioTTSService provides real-time text-to-speech using Fish Audio's high-performance WebSocket API. It's built for ultra-low latency conversational use cases.

Installation

pip install "piopiy-ai[fish]"

Prerequisites

• A Fish Audio API key (Get yours here).
• Set your API key in your environment:

  export FISH_API_KEY="your_api_key_here"

Configuration

FishAudioTTSService Parameters

Parameter	Type	Default	Description
api_key	str	Required	Your Fish Audio API key.
reference_id	str	Required	Reference ID of the voice model.
model_id	str	"s1"	Fish Audio model architecture.
params	InputParams	InputParams()	Generation settings.

InputParams

Parameter	Type	Default	Description
latency	str	"normal"	Latency mode (normal or balanced).
normalize	bool	True	Whether to normalize audio.
prosody_speed	float	1.0	Speed multiplier (0.5 to 2.0).
prosody_volume	int	0	Volume adjustment in dB.

Usage

Basic Setup

import os
from piopiy.services.fish.tts import FishAudioTTSService

tts = FishAudioTTSService(
    api_key=os.getenv("FISH_API_KEY"),
    reference_id="your_voice_reference_id"
)

Notes

• WebSocket Protocol: Uses MsgPack over WebSockets for efficient binary data transfer.
• Real-time Performance: Designed for ultra-low latency conversational use cases.
• Flush System: Supports explicit flushing to ensure all text is synthesized immediately.