local-audio example depends on openai module

[evmar]

Description

Environment

• pipecat-ai version: pipecat-ai==0.0.50
• python version: 3.12.4
• OS: MacOS 14.4.1

Issue description

When I tried running the local audio example I had a missing OpenAI dependency, despite that example code not mentioning OpenAI.

I have gathered there is some idea of "llm context" in the OpenAI module(?) but this example also doesn't seem to use llm context in its pipeline(?). (Sorry for getting terms wrong, just trying out this library for the first time.)

Repro steps

$ pip install "pipecat-ai[whisper]"
$ # copy of examples/foundational/01a-local-audio.py
$ python local.py  
[...]
2024-12-15 10:35:22.601 | ERROR    | pipecat.processors.aggregators.openai_llm_context:<module>:33 - Exception: No module named 'openai'
2024-12-15 10:35:22.601 | ERROR    | pipecat.processors.aggregators.openai_llm_context:<module>:34 - In order to use OpenAI, you need to `pip install pipecat-ai[openai]`. Also, set `OPENAI_API_KEY` environment variable.

Expected behavior

No dependency on OpenAI, or for the OpenAI dependency to be non-optional.

(I expected the above example to fail due to not installing the Cartesia support but that is not a bug.)

Actual behavior

Logs

[obigroup]

Hello,

Same issue with 01-say-one-thing.py, CartesiaTTSService requires openAI but I want to try the code with Anthropic.
How can I do that ?

Regards.

[aconchillo]

For 01a-local-audio.py you would need to do:

pip install "pipecat-ai[local,cartesia,openai]"

And for 01-say-one-thing.py

pip install "pipecat-ai[daily,cartesia,openai]"