Intentional - Pipecat#
Plugin that lets you build transcribed audio voice bots using Pipecat.
Table of Contents#
Installation#
You can install intentional-pipecat without extras:
pip install intentional-pipecat
However, it will be unusable as-is, because it will come with nearly no dependencies.
In order to have any of the underlying dependencies installed (such as Silero's VAD, Deepgram, OpenAI, etc...) you must specify them as extras, like:
pip install intentional-pipecat[silero,deepgram,openai]
For a list of all the available extras, see Pipecat's documentation or their pyproject.toml file. We are going to try to keep this list up-to-date on Intentional's end, but in case of issues you can also install those dependencies by hand by doing:
pip install pipecat-ai[<the extras you want>]
This guarantees that you get the correct extras for your version of Pipecat. Please open an issue if you find any discrepancies.
Usage#
In order to use Pipecat, you need to specify which components you want to use in the configuration file. For example:
bot:
type: pipecat
llm:
client: openai
name: gpt-4o
vad:
module: silero
client: SileroVADAnalyzer
stt:
module: deepgram
client: DeepgramSTTService
tts:
module: azure
client: AzureTTSService
This example would require the extras silero, deepgram and azure:
pip install intentional-pipecat[silero,deepgram,azure]
See Pipecat's documentation for more information about what modules and classes are available for the various pipeline components.
At this time, the Pipeline structure itself cannot be configured. Open an issue if you'd like to allow some degree of pipeline customization.
License#
intentional is distributed under the terms of the AGPL license. If that doesn't work for you, get in touch.