This repository demonstrates an Orchest pipeline that generates an audio fragment using the coqui TTS engine and sends it as a message on Slack
If you want to send this to your actual Slack channel you need a bot with chat:read/write and file:read/write permissions and set the SLACK_BOT_TOKEN or DISCORD_TOKEN environment variable.
Make sure you add the bot to the desired channel. The Slack channel can be overwritten by specifying the SLACK_CHANNEL environment variable.
- To use slack bot use
utils.send_slack_file. - To use discord bot use
utils.send_discord_file.
Check out the speaker_id_samples.zip to find out the difference between the various speaker_id options of the Coqui VITS model.
The VITS model that's used is based off of earlier work in the field. For an overview of the evolution check out the evolution trajectories below:
For more details check out the paper about recent advancements in neural speech synthesis: arxiv.org/abs/2106.15561