FireTTS από dishdare

Right-click to speak selected text using locally installed TTS (coqui-ai/TTS). Sends selected text to your offline TTS system for speech output. This is more or less DIY project since most of the things are needed on your local machine.

5 (1 reviews)

1 χρήστης

Θα χρειαστείτε το Firefox για να χρησιμοποιήσετε αυτήν την επέκταση

Λήψη Firefox και απόκτηση επέκτασης

Λήψη αρχείου

Σχετικά με την επέκταση

Text to Speech solution using locally installed TTS.

Repo - https://github.com/coqui-ai/TTS

steps to install coqui-ai/TTS locally
conda create --name tts python=3.11 jupyter -y
conda activate tts
pip install tts flask flask-cors
sudo apt install ffmpeg

create the server file and run it - server.py

server.py

    from flask import Flask, request
    from flask_cors import CORS
    import subprocess
    import os

    app = Flask(__name__)
    CORS(app)

    # Define output path
    OUTPUT_DIR = os.path.expanduser("~/tts")
    OUTPUT_FILE = os.path.join(OUTPUT_DIR, "output.wav")

    @app.route("/speak", methods=["POST"])
    def speak():
        text = request.json.get("text", "")
        print("Received text:", repr(text))

        if not text:
            return {"error": "No text provided"}, 400

        # Create output folder if it doesn't exist
        os.makedirs(OUTPUT_DIR, exist_ok=True)

        # Build and run the TTS command safely
        subprocess.run([
            "conda", "run", "-n", "tts",
            "tts",
            "--text", text,
            "--model_name", "tts_models/en/ljspeech/glow-tts",
            "--out_path", OUTPUT_FILE
        ])

        # Notify user
        os.system(f'notify-send "TTS" "Speaking: {text[:80]}..."')

        # Play the generated audio
        subprocess.run(["ffplay", "-nodisp", "-autoexit", OUTPUT_FILE])

        return {"status": "done"}

    if __name__ == "__main__":
        print("Starting server...")
        app.run(port=5002, debug=True)

Enjoy Freedom of Speech.

My first extension. Please let me know if there are any issues or solutions.
Special thanks to ChatGPT.

Βαθμολογήθηκε με 5 από έναν αξιολογητή