Teaching Your Raspberry Pi to Speak with IBM Watson

This tutorial shows how to make your Raspberry Pi talk using IBM Watson's Text-to-Speech service and Node-RED. We'll build on a previous example (assumed prior setup of Raspberry Pi, Bluemix, and Node-RED).

Artwork by SitePoint/PatCat, Logo rights: IBM and Raspberry Pi Foundation

Key Steps:

1. Enable Text-to-Speech in Bluemix: Add the "Text to Speech" service in your Bluemix dashboard. Note: This is not "Speech to Text." Choose a plan (e.g., "Standard"), name your service (e.g., "The Voice"), and create it. Record the credentials (username and password) from the "Service Credentials" section.

2. Install Node-RED Watson Nodes: SSH into your Raspberry Pi and navigate to your Node-RED directory (cd ~/.node-red). Install the necessary nodes using: sudo npm install node-red-node-watson

3. Install ALSA Library: Install the ALSA library for audio playback: sudo apt-get install libasound2-dev. Restart Node-RED (node-red-stop; node-red-start).

4. Create the Node-RED Flow: Add the following nodes to your Node-RED flow:

   • "text to speech" node (IBM_Watson): Configure with your Bluemix credentials, select a language and voice, and set the file format to WAV.

<code>*   **Function Node ("Temperature Text"):**  Format the temperature data (from a previous "getCPUTemp" node, assumed to exist) into a human-readable sentence:

```javascript
msg.payload = "My current CPU temperature is " + msg.payload.replace("temp=","").replace("'C\n","") + " degrees celsius";
return msg;
```

*   **Function Node ("Format Speech"):**  Prepare the speech message:

```javascript
msg.payload = msg.speech;
return msg;
```

*   **File Node ("Create WAV"):** Save the generated audio to `/home/pi/temp/speech.wav`.  Ensure "Create directory if it doesn’t exist" is checked and "Add newline" is unchecked.</code>

<code>*   **Delay Node:** Add a 1-second delay to allow the audio file to be created.

*   **File Node ("Read WAV"):** Read the audio file from `/home/pi/temp/speech.wav` as a buffer.

*   **Exec Node ("Play Audio"):** Play the audio using `aplay /home/pi/temp/speech.wav`.</code>

5. Configure Audio Output: Direct audio output to your 3.5mm jack (if needed) using amixer cset numid=3 1.

6. Deploy and Test: Deploy your Node-RED flow. Your Raspberry Pi should now speak!

Troubleshooting and FAQs: Refer to the original article's FAQ section for detailed troubleshooting and information on language support, voice customization, integration with other services, and limitations. Remember to consider usage limits and costs associated with the IBM Watson Text-to-Speech service.

The above is the detailed content of Teaching Your Raspberry Pi to Speak with IBM Watson. For more information, please follow other related articles on the PHP Chinese website!