chapter14.md

Chapter 14: Speech Recognition and Synthesis

You can use LangChain to create applications that incorporate speech recognition and speech synthesis.

from langchain.llms import OpenAI
from langchain.chains import LLMChain
from langchain.prompts import PromptTemplate
import speech_recognition as sr
import pyttsx3

# 
recognizer = sr.Recognizer()

# 
engine = pyttsx3.init()

# LLM
llm = OpenAI(temperature=0.7)
prompt = PromptTemplate(
    input_variables=["query"],
    template="You are a helpful assistant. Answer the following query: {query}"
)
chain = LLMChain(llm=llm, prompt=prompt)

# 
with sr.Microphone() as source:
    print("Speak your question:")
    audio = recognizer.listen(source)

try:
    # 
    query = recognizer.recognize_google(audio)
    print(f"You said: {query}")

    # LLM
    response = chain.run(query)
    print(f"AI response: {response}")

    # 
    engine.say(response)
    engine.runAndWait()

except sr.UnknownValueError:
    print("Sorry, I couldn't understand that.")
except sr.RequestError as e:
    print(f"Could not request results; {e}")