app.py

from fastapi import FastAPI
from pydantic import BaseModel
import boto3
import os
import base64

# aws Polly keys
polly = boto3.client(
    "polly",
    aws_access_key_id=os.environ["AWS_ACCESS_KEY_ID"],
    aws_secret_access_key=os.environ["AWS_SECRET_ACCESS_KEY"],
    region_name="eu-west-1"
)

app = FastAPI(title="Gaelic TTS")

class TTSRequest(BaseModel):
    text: str
    ipa: str
    rate: str = "75%"

#
@app.get("/")
def root():
    return {"status": "okay"}

# using mix of Gaelic text and IPA
def generate_tts(request: TTSRequest):
    ssml_text = f"""
    <speak>
        <prosody rate="{request.rate}">
            <phoneme alphabet="ipa" ph="{request.ipa}">{request.text}</phoneme>
        </prosody>
    </speak>
    """
    response = polly.synthesize_speech( 
        Text=ssml_text,
        TextType="ssml",
        OutputFormat="mp3",
        VoiceId="Gwyneth",
        Engine="standard"
    ) # welsh model
    audio_bytes = response["AudioStream"].read()
    return {"audio_base64": base64.b64encode(audio_bytes).decode("utf-8")}

@app.post("/predict") # call 
def predict(request: TTSRequest):
    return generate_tts(request)