Support for Local TTS function in the library

[Nikhil-punia]

Like the method bool connecttospeech(const char* speech, const char* lang) which works on the HTTP GET requests for text to speech task , there can be also a method for HTTP POST requests , that can be useful for :

• Sending more amount of data for TTS using POST
• Fast TTS
• Local TTS support

For e.g. : Here is a Function for handling the POST request :

bool Audio::connect_local_tts(const String& host, int port, const String& path,
                             const String& text, const String& voice_id,
                              const String& lang, const String& endpoint) {

    if (text.isEmpty()) {
        AUDIO_INFO("TTS text is empty");
        stopSong();
        return false;
    }

    xSemaphoreTakeRecursive(mutex_playAudioData, 0.3 * configTICK_RATE_HZ);
    setDefaults();
    m_f_ssl = false;
    m_f_tts = true;
    m_speechtxt.assign(text.c_str());

    String body = "{";
    body += "\"text\":\"" + text + "\"";
    if (!voice_id.isEmpty()) body += ",\"voice\":\"" + voice_id + "\"";
    if (!lang.isEmpty()) body += ",\"lang\":\"" + lang + "\"";
    body += "}";

    String req = "POST " + path + " HTTP/1.1\r\n";
    req += "Host: " + host + "\r\n";
    req += "User-Agent: ESP32-AudioI2S\r\n";
    req += "Accept: */*\r\n";
    req += "Content-Type: application/json\r\n";
    req += "Content-Length: " + String(body.length()) + "\r\n";
    req += "Connection: close\r\n\r\n";
    req += body;

    _client = static_cast<WiFiClient*>(&client);
    AUDIO_INFO("Connecting to local TTS server %s:%d", host.c_str(), port);

    if (!_client->connect(host.c_str(), port)) {
        AUDIO_ERROR("Connection failed");
        xSemaphoreGiveRecursive(mutex_playAudioData);
        return false;
    }

    _client->print(req);

    m_f_running = true;
    m_dataMode = HTTP_RESPONSE_HEADER;
    m_lastHost.assign(host.c_str());

    if (endpoint == "gtts" || endpoint == "edge_tts") {
        m_expectedCodec = CODEC_MP3;
    } else if (endpoint == "tts") {
        m_expectedCodec = CODEC_WAV;
    }

    xSemaphoreGiveRecursive(mutex_playAudioData);
    return true;
}

And this can be used with a simple Python Server running locally :

from flask import Flask, request, send_file, jsonify, after_this_request
import pyttsx3
import uuid
import os
from gtts import gTTS
import asyncio
import edge_tts

app = Flask(__name__)

# Load available pyttsx3 voices
engine = pyttsx3.init()
available_voices = {v.id: v.name for v in engine.getProperty('voices')}


@app.route('/voices', methods=['GET'])
def list_voices():
    """List available system voices (pyttsx3)."""
    return jsonify(available_voices)


@app.route('/tts', methods=['POST'])
def tts():
    """Offline TTS using pyttsx3."""
    data = request.get_json(force=True)
    text = data.get('text')
    voice_id = data.get('voice')
    rate = data.get('rate')

    if not text:
        return jsonify({"error": "Missing 'text' parameter"}), 400

    filename = f"/tmp/{uuid.uuid4().hex}.wav"

    try:
        engine = pyttsx3.init()
        if voice_id:
            engine.setProperty('voice', voice_id)
        if rate:
            engine.setProperty('rate', rate)
        engine.save_to_file(text, filename)
        engine.runAndWait()
    except Exception as e:
        return jsonify({"error": str(e)}), 500

    @after_this_request
    def remove_file(response):
        try:
            os.remove(filename)
        except Exception as e:
            app.logger.warning(f"Failed to delete file {filename}: {e}")
        return response

    return send_file(filename, mimetype='audio/wav')


@app.route('/gtts', methods=['POST'])
def gtts_endpoint():
    """Online TTS using Google gTTS."""
    data = request.get_json(force=True)
    text = data.get('text')
    lang = data.get('lang', 'en')

    if not text:
        return jsonify({"error": "Missing 'text' parameter"}), 400

    filename = f"/tmp/{uuid.uuid4().hex}.mp3"

    try:
        tts = gTTS(text=text, lang=lang)
        tts.save(filename)
    except Exception as e:
        return jsonify({"error": str(e)}), 500

    @after_this_request
    def remove_file(response):
        try:
            os.remove(filename)
        except Exception as e:
            app.logger.warning(f"Failed to delete file {filename}: {e}")
        return response

    return send_file(filename, mimetype='audio/mpeg')


@app.route('/edge_tts', methods=['POST'])
def edge_tts_route():
    """High-quality TTS using Microsoft Edge Neural voices (safe pitch/rate support)."""
    data = request.get_json(force=True)
    text = data.get('text')
    voice = data.get('voice', 'en-US-AriaNeural')
    rate = str(data.get('rate')) if 'rate' in data else None
    pitch = str(data.get('pitch')) if 'pitch' in data else None

    if not text:
        return jsonify({"error": "Missing 'text' parameter"}), 400

    filename = f"/tmp/{uuid.uuid4().hex}.mp3"

    async def generate_tts():
        options = {
            "text": text,
            "voice": voice
        }

        if rate and rate not in ["0%", "+0%", "-0%", "0", ""]:
            options["rate"] = rate
        if pitch and pitch not in ["0%", "+0%", "-0%", "0", ""]:
            options["pitch"] = pitch

        try:
            communicate = edge_tts.Communicate(**options)
            await communicate.save(filename)
        except Exception as e:
            if "pitch" in str(e).lower() or "rate" in str(e).lower():
                communicate = edge_tts.Communicate(text=text, voice=voice)
                await communicate.save(filename)
            else:
                raise e

    try:
        asyncio.run(generate_tts())
    except Exception as e:
        return jsonify({"error": str(e)}), 500

    return send_file(filename, mimetype='audio/mpeg')


if __name__ == '__main__':
    app.run(host='0.0.0.0', port=5000)

Thanks .

[schreibfaul1]

Thank you for your contribution.
I'm sure very few people will have a TTS server in their home network. But for those who have one or want to experiment with it, your code is a good start.
I have included your code in the examples so that it doesn't get lost in the mass of issues.
https://github.com/schreibfaul1/ESP32-audioI2S/tree/master/examples/Support%20for%20Local%20TTS%20function

[Nikhil-punia]

Thanks for the consideration and such a great library 🤗