feat: show fun preparation texts (while generating audio)

tts.py

@@ -2,6 +2,7 @@ import os
 from time import sleep, time
 import warnings
 import importlib
+import random
 
 import torch
 import argparse
@@ -19,6 +20,20 @@ voices = {
     'brit': 'bf_emma'
 }
 
+prep_texts = [
+    "Check mic, 1-2-3...",
+    "*Tap* *tap* ... Is this thing on?",
+    "Ready, set... *Ahem!*",
+    "Mic's on, lights are set, I'm ready to roll.",
+    "All set? Let's make it a good one.",
+    "Ready, set, go—now that's the real countdown.",
+    "Checking the mic, one, two, three.",
+    "Lights, mic, action—now let's do this.",
+    "Hold tight—this is about to get interesting.",
+    "If the mic works, we're good to go.",
+    "All systems green—let's make this a good one."
+]
+
 def parse_args():
     parser = argparse.ArgumentParser(description="Simple TTS", allow_abbrev=False)
     parser.add_argument(
@@ -83,7 +98,7 @@ def parse_args():
 
 def generate_audio(generator, name, voice):
     output_files = []
-    with yaspin():
+
     for i, (gs, ps, audio) in enumerate(generator):
         output_file_name=f'outputs/{name}/{name}-{voice}-{i}.wav'
         os.makedirs(os.path.dirname(output_file_name), exist_ok=True)
@@ -108,7 +123,16 @@ def play_audio(output_files):
             colour='yellow'):
             sleep(duration / 100)
 
+
 def main():
+
+    # Get a randome "preparing" text
+    spinner_text = random.choice(prep_texts)
+
+    # Generate audio
+    with yaspin() as spinner:
+        spinner.text = spinner_text
+ 
         args=parse_args()
 
         if not args.verbose:
@@ -172,6 +196,7 @@ def main():
             print(f"[TTS] {len(output_files)} chunks generated in {generation_time:.2f} seconds")
             print(f"[TTS] Output files are in: {directory}/*")
 
+    # Play audio
     if args.skip_play:
         print(f"[TTS] Audio player disabled: {directory}/*")
     else: