Kingston Kuan
50362ca37e
Merge branch 'pipecat-ai:main' into fix/smallwebrtcrequesthandler-return-type
2025-12-15 16:41:59 +08:00
Aleix Conchillo Flaqué
4f848e9631
Merge pull request #3227 from fixie-ai/mike/upstream
...
Add Ultravox service
2025-12-13 18:29:02 -08:00
Kingston
93df7044fa
fix return type for SmallWebRTCRequestHandler
2025-12-13 22:11:06 +08:00
Mike Depinet
2e4fa3f8db
PR comments
...
Also satisfy some Pyright complaints and update default model
2025-12-12 15:03:31 -08:00
Mark Backman
6cda357ce8
Remove timestamp check from TestThoughtTranscription
2025-12-12 14:28:39 -05:00
Mark Backman
7e87f61d17
Update GeminiLiveLLMService model to gemini-2.5-flash-native-audio-preview-12-2025
2025-12-12 14:18:57 -05:00
Mike Depinet
4b81be7acf
Add Ultravox service ( #1 )
...
Adds support for using Ultravox Realtime as a speech-to-speech service.
Also removes the deprecated Ultravox speech-to-text vllm model integration to avoid confusion.
2025-12-12 10:16:15 -08:00
Filipi Fuchter
b25ad21941
Refactoring HeyGenVideoService and HeyGenTransport to work with both APIs.
2025-12-12 08:51:35 -03:00
Filipi Fuchter
c2abe42a64
Adding support for the HeyGen LiveAvatar API.
2025-12-12 08:49:52 -03:00
Filipi Fuchter
56dee06a29
Refactored the Interactive Avatar API to extend the HeyGen base API.
2025-12-12 08:49:16 -03:00
Filipi Fuchter
60cc14cafd
Created HeyGen base API to support both Interactive Avatar and LiveAvatar.
2025-12-12 08:48:39 -03:00
kompfner
1e98094394
Merge pull request #3175 from pipecat-ai/pk/thinking-exploration
...
Additional functionality related to thinking, for Google and Anthropic LLMs.
2025-12-11 17:15:37 -05:00
Paul Kompfner
0e88ad672e
Add ThoughtTranscriptionMessage.role, which is always "assistant"
2025-12-11 14:41:16 -05:00
kompfner
f41c3dcbc3
Merge pull request #3212 from pipecat-ai/pk/nova-2-sonic
...
Nova 2 Sonic support
2025-12-11 09:36:50 -05:00
Mark Backman
645e1802f8
Merge pull request #3219 from pipecat-ai/mb/deprecate-fal-smart-turn
2025-12-10 13:13:44 -05:00
Mark Backman
6636da682c
Merge pull request #3085 from rimelabs/feature/rimeNonJsonTTsservice
...
Add RimeNonJsonTTSService for non-JSON WebSocket API support
2025-12-10 10:38:39 -05:00
Mark Backman
10a32c943f
deprecate: FalSmartTurnAnalyzer and LocalSmartTurnAnalyzer
2025-12-10 08:14:28 -05:00
Gokul Js
455579ffcc
Refactor RimeNonJsonTTSService to extend InterruptibleTTSService, removing dependency on WebsocketTTSService and streamlining audio interruption handling.
2025-12-10 04:56:52 +05:30
vipyne
ff2f1dac82
user-bot-latency log observer internal var change
2025-12-09 12:34:38 -06:00
Paul Kompfner
b821dd2507
Fix a bug in AWSNovaSonicLLMService where we would mishandle cancelled tool calls in context
2025-12-09 12:12:55 -05:00
Paul Kompfner
0c5bccd1f1
Changes related to Nova 2 Sonic's support for the model speaking first
2025-12-09 11:55:23 -05:00
Paul Kompfner
926514ca18
Add support to AWSNovaSonicLLMService for new "endpointingSensitivity" parameter.
2025-12-09 11:26:43 -05:00
Paul Kompfner
ca5e668f4a
Update AWSNovaSonicLLMService docstring with more (and more up-to-date) info
2025-12-09 10:14:27 -05:00
Paul Kompfner
b22ac8292f
Update default model in AWSNovaSonicLLMService to "amazon.nova-2-sonic-v1:0"
2025-12-09 09:38:47 -05:00
Gokul Js
2a6a0d83db
Update docstring in RimeNonJsonTTSService to clarify the focus on the current plain text protocol and note potential future support for JSON WebSocket.
2025-12-09 02:49:37 +05:30
Gokul Js
6ca117a3c1
Remove unused import of 'language' in tts.py to clean up the code and improve readability.
2025-12-09 02:45:17 +05:30
Aleix Conchillo Flaqué
88289f578a
Merge pull request #3208 from pipecat-ai/thor/add-client-identification
...
add Gemini client identification
2025-12-08 13:05:04 -08:00
Paul Kompfner
229ff794d6
Better handle Gemini non-function thought signatures
2025-12-08 15:56:40 -05:00
Aleix Conchillo Flaqué
096db3eb6c
Merge pull request #3207 from pipecat-ai/aleix/voicemail-conversation-detected-event
...
VoicemailDetector: add on_conversation_detected event
2025-12-08 11:59:45 -08:00
Aleix Conchillo Flaqué
cfd1cada8c
VoicemailDetector: add on_conversation_detected event
2025-12-08 11:57:14 -08:00
Aleix Conchillo Flaqué
d289b38ba7
tests(google): mock the new pipecat.version()
2025-12-08 11:51:01 -08:00
Aleix Conchillo Flaqué
b0f63c3785
pipecat: add version() function
2025-12-08 11:51:01 -08:00
Paul Kompfner
1249ee3de3
Better handle Gemini non-function thought signatures
2025-12-08 13:07:25 -05:00
vipyne
540a48b1b6
use VADUserStarted/StoppedSpeakingFrame s in user_bot_latency_log_observer.py
2025-12-08 11:37:31 -06:00
Paul Kompfner
aa0529ff82
Update comments for accuracy
2025-12-08 11:47:06 -05:00
Paul Kompfner
7e92597c0e
Remove LLMThoughtSignatureFrame in favor of using the more generic LLMMessagesAppendFrame
2025-12-08 11:10:05 -05:00
Gokul Js
0b4d984be6
Standardize error handling in RimeNonJsonTTSService by replacing specific error messages with a generic "Unknown error occurred" format, enhancing consistency in error reporting.
2025-12-08 21:24:30 +05:30
Paul Kompfner
17203ba3e6
Change FunctionInProgressFrame.llm_specific_extra to a more generic FunctionInProgressFrame.append_extra_context_messages.
2025-12-08 10:50:19 -05:00
Gokul Js
924831089c
Enhance error handling in RimeNonJsonTTSService by standardizing error messages for improved clarity and consistency in reporting.
2025-12-08 21:17:01 +05:30
Gokul Js
329b8ac426
Refactor error handling in RimeNonJsonTTSService to provide a more generic error message, improving clarity in error reporting.
2025-12-08 21:06:48 +05:30
Paul Kompfner
61674d7758
Add process_thought constructor argument to TranscriptProcessor to control whether to handle thoughts in addition to assistant utterances. Defaults to False.
2025-12-08 10:27:36 -05:00
Gokul Js
b9990811b5
Merge branch 'main' into feature/rimeNonJsonTTsservice
2025-12-08 20:54:01 +05:30
Gokul Js
f4e33fc8dd
Update docstrings in RimeNonJsonTTSService for clarity and consistency, specifying 'Non-JSON' in relevant descriptions.
2025-12-08 20:32:13 +05:30
Gokul Js
5bfea84bd5
Refactor RimeNonJsonTTSService to extend WebsocketTTSService, enhancing WebSocket functionality and improving code clarity
2025-12-08 20:30:46 +05:30
Paul Kompfner
ef703e9d16
Get rid of ThoughtTranscriptProcessor, moving its logic into AssistantTranscriptProcessor instead
2025-12-08 09:59:32 -05:00
Paul Kompfner
44aa11737b
Minor docstring update for accuracy
2025-12-08 09:29:10 -05:00
Paul Kompfner
4ea51ff67c
Slight refactor of handling thought-signature-containing special context messages in the Gemini adapter
2025-12-08 09:29:10 -05:00
Paul Kompfner
15f5583fd2
Simplify, at the expense of a bit of not-yet-needed flexibility: rather than associating a loose thought_metadata with each thought, use a signature. Thought signatures are the only "thought metadata" we use today.
2025-12-08 09:29:10 -05:00
Paul Kompfner
c8c6f424cd
Add support for Gemini 3 Pro non-function-call-related thought signatures
2025-12-08 09:29:10 -05:00
Paul Kompfner
217f03b9cc
Add additional functionality related to "thinking", for Google and Anthropic LLMs.
...
Thinking, sometimes called "extended thinking" or "reasoning", is an LLM process where the model takes some additional time before giving an answer. It's useful for complex tasks that may require some level of planning and structured, step-by-step reasoning. The model can output its thoughts (or thought summaries, depending on the model) in addition to the answer. The thoughts are usually pretty granular and not really suitable for being spoken out loud in a conversation, but can be useful for logging or prompt debugging.
Here's what's added:
1. New typed input parameters for Google and Anthropic LLMs that control the models' thinking behavior (like how much thinking to do, and whether to output thoughts or thought summaries).
2. New frames for representing thoughts output by LLMs.
3. A generic mechanism for associating extra LLM-specific data with a function call in context, used specifically to support Google's function-call-related "thought signatures", which are necessary to ensure thinking continuity between function calls in a chain (where the model thinks, makes a function call, thinks some more, etc.)
4. A generic mechanism for recording LLM thoughts to context, used specifically to support Anthropic, whose thought signatures are expected to appear alongside the text of the thoughts within assistant context messages.
5. An expansion of `TranscriptProcessor` to process LLM thoughts in addition to user and assistant utterances.
2025-12-08 09:29:01 -05:00
