These scripts translate speech input, and output the translated speech on another computer. The client script records the voice to a WAV file, transcripts the voice to text using Google Speech Translation. The text is then translated with the Googletrans python module, and the translated text is transferred to another computer with speakers. The server script on the other computer converts the received text to mp3, and opens the mp3 file.
The idea for this project appeared following Spanish lessons one evening. I imagined a future with a raspberry pi, a headset and a speaker duct taped to my shoulder. I would speak Spanish into the headset, the pi would translate and output the translations through the speaker. I have both a headset and a pi, however I have no usb speaker. As an improvisation, I thought instead that my translation could output from my lounge room computer speakers.
Add the client.py and config.json file to the raspberry pi, or computer that will receive your microphone voice input. Add the server.py and config.json to the computer that will receive the translated text and will output the translated audio file.
Supports linux/windows and python3.
You will need to install the following python modules:
- pyaudio
- wave
- json
- SpeechRecognition
- Googletrans
- gTTS
Firstly, run the client.py script on the receiving computer. The receiving computer will wait to receive the translated text. On the sending computer, run the server.py script and indicate how many seconds to voice record. For example, to record for five seconds run the following command:
python client.py 5
When the script starts, you it will indicate when it starts and stops recording.
After you have recorded the audio, the receiving computer will output the text received and then open the default program you use to open mp3s.
I would like to add voice control to edit the options. The options available would be the record time and language destination/source (i.e. English to Spanish). I'd like generate the output audio on the client script instead of the server script. Finally, I'd like create another version with a single script which outputs through a pi usb speaker (duct taped to my shoulder).