Revert "use coqui tts instead of mimic"

This reverts commit c325154.

Author: texhnolyze

bitbots_misc/bitbots_tts/bitbots_tts/tts.py

@@ -1,24 +1,23 @@
 #!/usr/bin/env python3
 
+import os
+import subprocess
+import time
 import traceback
 
 import rclpy
-import sounddevice as sd
+import requests
+from ament_index_python import get_package_prefix
 from rcl_interfaces.msg import Parameter, SetParametersResult
 from rclpy.callback_groups import MutuallyExclusiveCallbackGroup
 from rclpy.executors import MultiThreadedExecutor
 from rclpy.node import Node
 from rclpy.publisher import Publisher
-from TTS.api import TTS
 
 from bitbots_msgs.msg import Audio
 
-# pip install TTS needed
 
-
-def speak(
-    text: str, publisher: Publisher, priority: int = 20, speaking_active: bool = True
-) -> None:  # Needed for what?
+def speak(text: str, publisher: Publisher, priority: int = 20, speaking_active: bool = True) -> None:
     """Utility method which can be used by other classes to easily publish a message."""
     if speaking_active:
         msg = Audio()
@@ -27,6 +26,13 @@ def speak(
         publisher.publish(msg)
 
 
+def say(text: str) -> None:
+    """Start the shell `say.sh` script to output given text with mimic3. Beware: this is blocking."""
+    script_path = os.path.join(get_package_prefix("bitbots_tts"), "lib/bitbots_tts/say.sh")
+    process = subprocess.Popen((script_path, text))
+    process.wait()
+
+
 class Speaker(Node):
     """
     Uses tts to say messages from the speak topic.
@@ -38,6 +44,9 @@ def __init__(self) -> None:
 
         # Class Variables
         self.prio_queue: list[tuple[str, int]] = []
+        self.speak_enabled = None
+        self.print_say = None
+        self.message_level = None
 
         # Initialize Parameters
         self.declare_parameter("print", True)
@@ -53,7 +62,16 @@ def __init__(self) -> None:
         # Subscribe to the speak topic
         self.create_subscription(Audio, "speak", self.speak_cb, 10, callback_group=MutuallyExclusiveCallbackGroup())
 
-        self.tts = TTS("tts_models/en/ek1/tacotron2").to("cpu")
+        # Wait for the mimic server to start
+        while True:
+            try:
+                requests.get("http://localhost:59125")
+                break
+            except requests.exceptions.ConnectionError:
+                # log once per second that the server is not yet available
+                self.get_logger().info("Waiting for mimic server to start...", throttle_duration_sec=2.0)
+                time.sleep(0.5)
+                pass
 
         # Start processing the queue
         self.create_timer(0.1, self.run_speaker, callback_group=MutuallyExclusiveCallbackGroup())
@@ -78,7 +96,7 @@ def run_speaker(self) -> None:
             # Get the next message and speak it
             text, _ = self.prio_queue.pop(0)
             try:
-                self.say(text)
+                say(text)
             except OSError:
                 self.get_logger().error(str(traceback.format_exc()))
 
@@ -99,10 +117,6 @@ def speak_cb(self, msg: Audio) -> None:
             self.prio_queue.append((text, prio))
             self.prio_queue.sort(key=lambda x: x[1], reverse=True)
 
-    def say(self, text: str) -> None:
-        sd.play(self.tts.tts_to_file(text, "test.wav"))
-        sd.wait()
-
 
 def main(args=None):
     rclpy.init(args=args)

bitbots_misc/bitbots_tts/launch/tts.launch

@@ -1,4 +1,7 @@
 <launch>
+    <!-- Do not launch mimic3-server on robot, as it is running as a systemd service -->
+    <executable unless="$(env IS_ROBOT)" cmd="mimic3-server --preload-voice en_US/vctk_low --preload-voice en_UK/apope_low --cache-dir $(env HOME)/.cache/mimic3/" name="mimic3-server" output="screen"/>
+
     <node name="bitbots_tts" pkg="bitbots_tts" exec="tts">
         <param from="$(find-pkg-share bitbots_tts)/config/tts_config.yaml"/>
     </node>