This repository packages Bark as a Truss.

Bark is a generative audio model for text-to-speech generation.

First, clone this repository:

git clone https://github.com/basetenlabs/truss-examples/
cd bark-truss

Before deployment:

1. Make sure you have a Baseten account and API key.
2. Install the latest version of Truss: pip install --upgrade truss

With bark-truss as your working directory, you can deploy the model with:

truss push

Paste your Baseten API key if prompted.

For more information, see Truss documentation.

Bark takes a string as its input and returns a Base64-encoded WAV audio as output. Bark currently works best for strings resulting in up to 12 seconds of audio, or approximately 20-25 words in English.

Here's an example invocation that decodes and saves the output to a file:

truss predict -d '"Two elevator mechanics discuss everything they hate about escalators"' | python process.py

With process.py as follows:

import base64
import sys

b64_audio = sys.stdin.read()

wav_file = open("bark_output.wav", "wb")
decode_string = base64.b64decode(b64_audio)
wav_file.write(decode_string)