Mark Backman comments

Results 240 comments of


                                            Mark Backman

input_tokens could be None when calculate ussage (Anthropic service）

In what scenario does this happen? It would be helpful if you add repro steps.

vertex ai on user_idle.push_frame(LLMMessagesFrame(messages)) giving error

Which LLM were you using?

Add function calling and device controls to the simple chatbot example

@mattieruth is this PR still relevant?

Lagging latency in Aws Nova sonic exmaple

Tagging @kompfner to take a look.

AI Assist to build with PipeCat

This is great!

Bot Audio Out Cracking and Missing Frames

Is anyone still experiencing this issue? I've run with ElevenLabs a lot and haven't hit it. If you are hitting it, can you share more about the machine you're running...

Bot Audio Out Cracking and Missing Frames

Closing this issue due to inactivity. Reopen if you're still impacted.

[Feature request] Groq transcriptions support

Added here: https://github.com/pipecat-ai/pipecat/pull/1183

[Feature request] Groq transcriptions support

This has been merged!

question: Is it possible to send audio only to one or more participant (not the entire mix) ?

You can't send the audio to specific participants, but you can selectively subscribe to participants. For example, in the translation demo, user A has speech translated. They don't need to...