piper_ros

This repository provides a set of ROS 2 packages to integrate piper TTS (Text-to-Speech) into ROS 2 using audio_common.

ROS 2 Distro	Branch	Build status	Docker Image
Humble	main
Iron	main
Jazzy	main
Kilted	main
Rolling	main

Table of Contents

1. Installation
2. Docker
3. Usage
4. Params

Installation

To run piper_ros follow the next commands:

cd ~/ros2_ws/src
git clone https://github.com/mgonzs13/piper_ros.git
cd ~/ros2_ws
vcs import src < src/whisper_ros/dependencies.repos
rosdep install --from-paths src --ignore-src -r -y
colcon build

Docker

You can build the piper_ros docker:

docker build -t piper_ros .

Then, you can run the docker container:

docker run -it --rm --device /dev/snd piper_ros

Usage

ros2 launch piper_bringup piper.launch.py

ros2 action send_goal /say audio_common_msgs/action/TTS "{'text': 'Hello World from ros 2'}"

Spanish Example

ros2 launch piper_bringup piper_spanish.launch.py

ros2 action send_goal /say audio_common_msgs/action/TTS "{'text': 'Hola Mundo desde ros 2'}"

Params

General Parameters

Param	Type	Default	Description
chunk	int32	512	Chunk size in samples for audio publication.
frame_id	string	""	Frame ID attached to published AudioStamped headers.

Model Parameters (model.*)

Param	Type	Default	Description
model.repo	string	"rhasspy/piper-voices"	HuggingFace repository for model download.
model.filename	string	"en/en_US/lessac/low/en_US-lessac-low.onnx"	Filename of the model in the repository.
model.path	string	""	Local path to the voice model file. If empty, the model is downloaded from model.repo.
model.config_repo	string	"rhasspy/piper-voices"	HuggingFace repository for the model config download.
model.config_filename	string	"en/en_US/lessac/low/en_US-lessac-low.onnx.json"	Filename of the model config in the repository.
model.config_path	string	""	Local path to the JSON voice config file. If empty, the config is downloaded from model.config_repo.

Synthesis Parameters (synthesis.*)

Param	Type	Default	Description
synthesis.speaker_id	int32	0	Numerical speaker ID for multi-speaker voices.
synthesis.noise_scale	float	0.667	Amount of noise added during audio generation.
synthesis.length_scale	float	1.0	Speed of speaking (1 = normal, < 1 faster, > 1 slower).
synthesis.noise_w_scale	float	0.8	Variation in phoneme lengths during synthesis.
synthesis.sentence_silence_seconds	float	0.2	Seconds of silence inserted between sentences.