Custom wake-up response

1. Course Content

• Adding audio files to the multi_brains program audio library allows you to customize the wake-up voice response.

[!NOTE]

• This tutorial is only for users who need to customize personalized voice responses and does not affect normal use.
• If you do not need to customize personalized responses, you can skip this section.

2. Preparing Audio Files

Note: You need to copy the generate_voice file from the "Source Code" folder to the current board for testing. For Jetson Nano and Raspberry Pi 5 boards, you will need to use Docker.

• First, place the generate_vioce file from the source code in the root directory. Then, place the command in the following path:

xxxxxxxxxx
sudo cp generate_vioce /usr/local/bin/

• You can prepare the audio material for the voice reply yourself.
• Alternatively, you can generate the voice using the system's built-in generate_vioce command. The voice generation model uses the Alibaba Cloud Model Studio Platform's speech synthesis model, so you need to configure ALIYUN_API_KEY according to the tutorial in 02-Configure API-KEY of this chapter first.
• Pre-configure Alibaba Cloud Model Studio Platform API-KEY

• Run the command in the terminal:

xxxxxxxxxx
generate_vioce --text Hello --language_type English

• Where --text is the startup parameter, "Hello" Replace with the content you want to synthesize the speech for.

• Audio files will be automatically saved in the ~/generate_voices/ directory.

[!NOTE] Other optional startup parameters are as follows:

• --voice: Speaker, default Cherry
• --language_type: Language, specify the language of the synthesized audio. Available options include Chinese, English, German, Italian, Portuguese, Spanish,Japanese, Korean, French, and Russian.
• --save_path: Audio file save path, default ~/
• --config_file: Configuration file path, default~/yahboomcar_ros2_ws/yahboomcar_ws/src/multi_brains/config/multi_brains_setting.yaml
• --text: Text to be synthesized audio, default is empty
• --model The speech synthesis model defaults to qwen3-tts-flash.

For available speakers and speech synthesis models, please refer to the dynamic notifications on the Bailian official website: https://bailian.console.aliyun.com/?spm=5176.29619931.J_SEsSjsNv72yRuRFS2VknO.2.74cd10d73l2Pw5&tab=doc#/doc/?type=model&url=2879134

• Reference Model:

• Reference Voice

• Supported Languages

Chinese, English, Spanish, Russian, Italian, French, Korean, Japanese, German, Portuguese

3. Loading Audio Files

• Multi_brains system audio path:

xxxxxxxxxx
/home/jetson/

• Where:
• zh is the Chinese response audio, suitable for Chinese users. Place the prepared audio files in this directory:

xxxxxxxxxx
~/yahboomcar_ros2_ws/yahboomcar_ws/src/multi_brains/system_vioce/zh/wakeup

• en is the English response audio, suitable for international users. Place the prepared audio files in this directory:

xxxxxxxxxx
~/yahboomcar_ros2_ws/yahboomcar_ws/src/multi_brains/system_vioce/en/wakeup

• When the multi_brains program is launched, the audio file in the corresponding directory will be automatically loaded, and a personalized response voice will be randomly played when the user wakes up.