43 lines
1.4 KiB
Python
43 lines
1.4 KiB
Python
# Theoretical Upgrade Test
|
|
from piper.voice import PiperVoice # Backbone of text to speech
|
|
import pyaudio
|
|
import subprocess
|
|
import os
|
|
import time
|
|
|
|
result = subprocess.run(["xsel"], capture_output=True, text=True) # In Theory grab selected text into Varible called TextToSpeak
|
|
time.sleep(1)
|
|
#print(result) # In Theory grab selected text and print ot to terminal
|
|
|
|
texttospeak = str(result.stdout)
|
|
print(texttospeak)
|
|
|
|
# texttospeak = "Welcome to the world of speech synthesis! This is an example of text-to-speech using Piper TTS."
|
|
|
|
#voice = PiperVoice.load("~/en_US-amy-medium.onnx") # Load the model
|
|
|
|
voicedir = os.path.expanduser('~') #Where onnx model files are stored on my machine
|
|
model = voicedir+"/en_GB-alba-medium.onnx"
|
|
voice = PiperVoice.load(model)
|
|
|
|
chunks = voice.synthesize(texttospeak) # Set up Piper to stream audio from TextToSpeak
|
|
first_chunk = next(chunks) # Get the first chunk to set up audio stream configuration
|
|
|
|
p = pyaudio.PyAudio()
|
|
stream = p.open( # Open audio stream with correct settings
|
|
format=p.get_format_from_width(first_chunk.sample_width),
|
|
channels=first_chunk.sample_channels,
|
|
rate=first_chunk.sample_rate,
|
|
output=True,
|
|
)
|
|
|
|
stream.write(first_chunk.audio_int16_bytes) # Play the first chunk
|
|
|
|
for chunk in chunks: # Play subsequent chunks of audio
|
|
stream.write(chunk.audio_int16_bytes)
|
|
input()
|
|
|
|
# Cleanup
|
|
stream.stop_stream()
|
|
stream.close()
|
|
p.terminate() |