Text-to-Speech on the Command Line

Edit I have found a new way to do this locally that gives excellent results. You can read the guide here /Edit

I’ve been practicing my Spanish recently, and I wanted a tool to output spoken Spanish from a text file as I read along.

I tried a few tools like espeak-NG, pyttsx3 and Festival but the results were a little robotic.

The tool that produced the best results was gTTS. This tool sends written text to Google Translate’s text-to-speech API and returns the text in spoken audio. There are many languages available (get the full list with gtts-cli --all) and the audio out can either be saved to a file or sent to stdout.

I want to the audio immediately, so I’m going to send the audio to standard out and pipe it into an audio player that will accept input on stdin; ffplay, provided by the ffmpeg package.

Here is the command I used:

gtts-cli --lang es --file written-text.txt | ffplay -nodisp -

• gtts-cli - The command line invocation of gTTS
• --lang es - The language to read the text in
• --file written-text.txt - The input text file
• ffplay - The audio player
• -nodisp - Disable the graphical display
• - - Accept the audio from stdin

If you want to save the audio to a file then append --output <file> to the gtts-cli command e.g.:

gtts-cli --lang es --file written-text.txt --output spoken-text.mp3

Keep in mind that this tool needs an internet connection and will send your data to Google.