bob_coquitts.coquitts_node

Attributes

DEFAULT_MODEL_NAME
DEFAULT_SAMPLE_RATE

Classes

RedirectOutput	Context manager to redirect stdout and stderr to a logger.
CoquiTTSnode	A ROS 2 node for text-to-speech synthesis using the Coqui TTS library.

Functions

main([args])

Main entry point for the ROS node.

Module Contents

bob_coquitts.coquitts_node.DEFAULT_MODEL_NAME = 'tts_models/en/ljspeech/vits'

bob_coquitts.coquitts_node.DEFAULT_SAMPLE_RATE = 24000

class bob_coquitts.coquitts_node.RedirectOutput(logger, level='info')

Context manager to redirect stdout and stderr to a logger.

This is a workaround to capture and redirect the verbose print statements from the Coqui TTS library into the ROS logger, keeping the console clean.

_logger

_log_level = 'info'

_buffer

__enter__()

__exit__(exc_type, exc_val, exc_tb)

class bob_coquitts.coquitts_node.CoquiTTSnode

Bases: rclpy.node.Node

A ROS 2 node for text-to-speech synthesis using the Coqui TTS library.

This node subscribes to a ‘text’ topic, buffers incoming text fragments, and intelligently splits them into complete sentences. These sentences are then processed in configurable chunks to generate audio. It offers extensive ROS parameters to control the TTS model, voice cloning, audio playback, and text processing behavior like filtering and trimming. It also publishes the text being synthesized to a ‘text_speaking’ topic.

sub

speaking_pub

text_buffer = ''

flush_timer = None

_get_unique_filepath(filepath: pathlib.Path) → pathlib.Path

Generates a unique filepath by appending a counter if the file already exists.

Args:

filepath (pathlib.Path): The desired initial filepath.

Returns:

pathlib.Path: The original filepath if it’s unique, otherwise a new

unique filepath with a numbered suffix.

text_callback(msg)

Callback for the /text topic; appends text and manages a flush timer.

This function’s sole responsibility is to append incoming raw text chunks to the internal buffer and reset a timer. All complex processing is deferred to the flush_buffer_callback to ensure that text is only processed after a natural pause in the input stream.

Args:

msg (std_msgs.msg.String): The message containing the text chunk.

flush_buffer_callback()

Normalizes, splits, and processes the entire text buffer for TTS.

This method is the core of the text processing pipeline, triggered by a timer after a pause in incoming text. It performs the following steps: 1. Takes the entire accumulated raw text from the buffer. 2. Applies the full normalization suite (regex filter, character filter,

and number separator removal) to the complete text block.

3. Based on the split_sentences parameter, it either: a) (Default) Splits the cleaned text into sentences using custom

   delimiters and processes them in chunks.

   2. Hands the entire cleaned text block to the TTS engine, relying on Coqui’s internal splitter.

_process_text(text_to_process: str)

Handles the final synthesis of a given text chunk.

This function receives a fully prepared string (either a manually created sentence chunk or a complete text block for Coqui to split). It performs a final strip of leading/trailing characters, publishes the text, and calls the Coqui TTS engine to generate and play/save the audio.

Args:

text_to_process (str): The text string to synthesize.

bob_coquitts.coquitts_node.main(args=None)

Main entry point for the ROS node.