How to create a voice translation bot with witai

In our globalized world, communication across linguistic boundaries is more essential than ever. In this article, we will explore how to implement this technology to make communication more inclusive and accessible to everyone.

The code is available here
on my github

First thing to do is install the dependencies

blinker==1.8.2
cachetools==5.5.0
certifi==2024.8.30
chardet==3.0.4
charset-normalizer==3.4.0
click==8.1.7
colorama==0.4.6
Flask==3.0.3
google-api-core==2.22.0
google-auth==2.36.0
google-cloud-texttospeech==2.21.0
googleapis-common-protos==1.65.0
googletrans==4.0.0rc1
grpcio==1.67.1
grpcio-status==1.67.1
gTTS==2.5.3
h11==0.9.0
h2==3.2.0
hpack==3.0.0
hstspreload==2024.11.1
httpcore==0.9.1
httpx==0.13.3
hyperframe==5.2.0
idna==2.10
itsdangerous==2.2.0
Jinja2==3.1.4
Levenshtein==0.26.1
MarkupSafe==3.0.2
playsound==1.2.2
prompt_toolkit==3.0.48
proto-plus==1.25.0
protobuf==5.28.3
pyasn1==0.6.1
pyasn1_modules==0.4.1
PyAudio==0.2.14
python-Levenshtein==0.26.1
RapidFuzz==3.10.1
requests==2.32.3
rfc3986==1.5.0
rsa==4.9
sniffio==1.3.1
SpeechRecognition==3.11.0
typing_extensions==4.12.2
urllib3==2.2.3
wcwidth==0.2.13
Werkzeug==3.1.2
wit==6.0.1

Audio to text conversion

from gtts import gTTS
import playsound
import os

def speak_translation(text, lang):
    tts = gTTS(text=text, lang=lang)
    filename = "translation.mp3"
    tts.save(filename)
    playsound.playsound(filename)
    os.remove(filename)

Google cloud text Speech

from google.cloud import texttospeech

def synthesize_speech(text, language_code="wo-WO", voice_name="wo-WO-Standard-A", output_file="output.mp3"):
    client = texttospeech.TextToSpeechClient()

    input_text = texttospeech.SynthesisInput(text=text)

    # Configurez la voix pour le Wolof
    voice = texttospeech.VoiceSelectionParams(
        language_code=language_code,
        name=voice_name,
        ssml_gender=texttospeech.SsmlVoiceGender.NEUTRAL,
    )

    # Paramètres audio
    audio_config = texttospeech.AudioConfig(
        audio_encoding=texttospeech.AudioEncoding.MP3
    )

    # Synthèse vocale
    response = client.synthesize_speech(
        input=input_text, voice=voice, audio_config=audio_config
    )

    # Sauvegarder le fichier audio
    with open(output_file, "wb") as out:
        out.write(response.audio_content)
        print(f"Audio content written to file {output_file}")

# Utilisez cette fonction avec votre texte
synthesize_speech("Bonjour, je teste la traduction en Wolof.", "wo-WO")

Translation

from googletrans import Translator

def translate_text(text, target_lang):
    try:
        translator = Translator()
        translation = translator.translate(text, dest=target_lang)
        print(f"Traduction : {translation.text}")
        return translation.text
    except Exception as e:
        print(f"Erreur lors de la traduction : {e}")
        return "Traduction non disponible"

Voice detection

import speech_recognition as sr

def record_audio():
    recognizer = sr.Recognizer()
    with sr.Microphone() as source:
        print("Parlez maintenant...")
        audio = recognizer.listen(source)
        try:
            text = recognizer.recognize_google(audio, language="fr-FR")
            print(f"Vous avez dit : {text}")
            return text
        except sr.UnknownValueError:
            print("Désolé, je n'ai pas compris.")
        except sr.RequestError as e:
            print(f"Erreur de service : {e}")

Witai params:
You must go to the Meta API (Facebook) to create your token

import requests

WIT_AI_TOKEN = 'xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx'

def send_to_wit(text):
    headers = {'Authorization': f'Bearer {WIT_AI_TOKEN}'}
    response = requests.get(f'https://api.wit.ai/message?v=20230414&q={text}', headers=headers)
    return response.json()

The main file

from flask import Flask, request, jsonify
from convertion_audio_to_text import speak_translation
from translation import translate_text
from voice_detection import record_audio
from witai_params import send_to_wit
import Levenshtein

app = Flask(__name__)

# Langues disponibles
AVAILABLE_LANGUAGES = {
    "sw": "Swahili",
    "wo": "Wolof",
    "fon": "Fon",
    "en": "Anglais",
    "fr": "Français"
}

def calculate_score(reference_text, user_text):
    similarity = Levenshtein.ratio(reference_text.lower(), user_text.lower()) * 100
    return round(similarity, 2)

@app.route('/available_languages', methods=['GET'])
def available_languages():
    """Retourne les langues disponibles pour la traduction."""
    return jsonify(AVAILABLE_LANGUAGES)


@app.route('/process_audio', methods=['POST'])
def process_audio():
    """Traite l'audio, traduit le texte et évalue la prononciation."""
    try:
        # Étape 1 : Récupérer la langue cible depuis la requête
        target_lang = request.json.get('target_lang')

        if not target_lang:
            return jsonify({"error": "Paramètre 'target_lang' manquant"}), 400

        if target_lang not in AVAILABLE_LANGUAGES:
            return jsonify({
                "error": f"Langue cible '{target_lang}' non supportée.",
                "available_languages": AVAILABLE_LANGUAGES  # Retourner la liste des langues disponibles
            }), 400

        # Étape 2 : Traduire le texte initial
        text = record_audio()
        if not text:
            return jsonify({"error": "No audio detected or transcription failed"}), 400

        wit_response = send_to_wit(text)
        print("Wit.ai Response:", wit_response)

        translation = translate_text(text, target_lang)
        speak_translation(translation, lang=target_lang)

        # Étape 3 : Boucle de répétition pour évaluer la prononciation
        score = 0
        while score < 80:
            repeat_text = record_audio()
            if not repeat_text:
                return jsonify({"error": "No repeated audio detected"}), 400

            score = calculate_score(translation, repeat_text)
            if score >= 80:
                message = "Bravo! Félicitations, vous êtes un génie!"
                return jsonify({
                    "original_text": text,
                    "wit_response": wit_response,
                    "translated_text": translation,
                    "repeated_text": repeat_text,
                    "score": score,
                    "message": message
                }), 200
            elif score < 45:
                message = "Votre score est faible, améliorez-vous en vous entraînant."
            else:
                message = "Pas mal! Vous pouvez encore améliorer."

            return jsonify({
                "translated_text": translation,
                "repeated_text": repeat_text,
                "score": score,
                "message": message,
                "retry": True
            })

    except Exception as e:
        return jsonify({"error": str(e)}), 500


if __name__ == '__main__':
    app.run(debug=True)


"""
tu peux tester avec ce code dans le navigateur, tu decommente, puis tu le met la ou il faut
@app.route('/process_audio', methods=['GET', 'POST'])
def process_audio():
    if request.method == 'GET':
        return jsonify({"message": "Utilisez une requête POST pour traiter l'audio."})

    # Continue avec la logique POST
    try:
        text = record_audio()
        if not text:
            return jsonify({"error": "No audio detected or transcription failed"}), 400

        wit_response = send_to_wit(text)
        print("Wit.ai Response:", wit_response)

        target_lang = request.json.get('target_lang', 'sw')
        translation = translate_text(text, target_lang)

        speak_translation(translation, lang=target_lang)

        return jsonify({
            "original_text": text,
            "wit_response": wit_response,
            "translated_text": translation
        }), 200
    except Exception as e:
        return jsonify({"error": str(e)}), 500
"""

Designing a bot is becoming easier and easier today to solve complex problems in our daily lives. However, this does not exclude the importance of learning languages ​​on your own. The use of technologies like BotAI for instant voice translation should primarily serve to enrich our interactions in complex contexts. By combining these tools with personal language learning, we promote more effective communication while promoting individual linguistic wealth.

The code is available here
on my github

The above is the detailed content of How to create a voice translation bot with witai. For more information, please follow other related articles on the PHP Chinese website!