Paul Kompfner
75b3ea9c96
Update OpenAIRealtimeLLMService to work with LLMContext and LLMContextAggregatorPair (cont'd).
...
Fix tracing.
2025-10-29 15:43:51 -04:00
Paul Kompfner
95be1510ac
Update OpenAIRealtimeLLMService to work with LLMContext and LLMContextAggregatorPair (cont'd).
...
Improve `OpenAIRealtimeLLMAdapter.get_messages_for_logging()`.
2025-10-29 15:43:51 -04:00
Paul Kompfner
df19011080
Update OpenAIRealtimeLLMService to work with LLMContext and LLMContextAggregatorPair (cont'd).
...
Improve warning about transcription frame direction change.
2025-10-29 15:43:51 -04:00
Paul Kompfner
e42cf78e79
Update OpenAIRealtimeLLMService to work with LLMContext and LLMContextAggregatorPair (cont'd).
...
Update deprecation versions.
2025-10-29 15:43:51 -04:00
Paul Kompfner
0495de52b6
Update OpenAIRealtimeLLMService to work with LLMContext and LLMContextAggregatorPair (cont'd).
...
Log warning about transcription frame direction change.
2025-10-29 15:43:51 -04:00
Paul Kompfner
9bc02afd0d
Update OpenAIRealtimeLLMService to work with LLMContext and LLMContextAggregatorPair (cont'd).
...
CHANGELOG tweak.
2025-10-29 15:43:51 -04:00
Paul Kompfner
6140fdb2c9
Update OpenAIRealtimeLLMService to work with LLMContext and LLMContextAggregatorPair (cont'd).
...
In anticipation of `messages` property being added to `LLMContext` (in another PR), remove warnings about the need to use `get_messages()` instead.
2025-10-29 15:43:51 -04:00
Paul Kompfner
b6a1886dae
Update OpenAIRealtimeLLMService to work with LLMContext and LLMContextAggregatorPair (cont'd).
2025-10-29 15:43:51 -04:00
Paul Kompfner
42d0a097c5
Tweaks to 20b example
2025-10-29 15:43:51 -04:00
Paul Kompfner
3761804146
Make OpenAIRealtimeLLMService's websocket send method more resilient. Previously, it was possible for a websocket send attempt to occur during a disconnect.
2025-10-29 15:43:51 -04:00
Paul Kompfner
46e97c57c2
Update OpenAIRealtimeLLMService to work with LLMContext and LLMContextAggregatorPair (cont'd).
...
Update 20b example to use new `LLMContext` pattern.
2025-10-29 15:43:51 -04:00
Paul Kompfner
19770b76b4
Update OpenAIRealtimeLLMService to work with LLMContext and LLMContextAggregatorPair (cont'd).
...
Add back file that was removed, when it should've just been deprecated.
Also, fix version numbers in deprecation messages to match the next expected release.
2025-10-29 15:43:51 -04:00
Paul Kompfner
b34461bf93
Update OpenAIRealtimeLLMService to work with LLMContext and LLMContextAggregatorPair (cont'd).
2025-10-29 15:43:47 -04:00
Paul Kompfner
bab0aaf585
Update OpenAIRealtimeLLMService to work with LLMContext and LLMContextAggregatorPair (cont'd).
...
Update `create_context_aggregator()` (which we're keeping around for backward compatibility) to create a `LLMContextAggregatorPair` rather than OpenAI-Realtime-specific aggregators.
2025-10-29 15:36:58 -04:00
Paul Kompfner
61944d22ef
Update OpenAIRealtimeLLMService to work with LLMContext and LLMContextAggregatorPair (cont'd).
...
Implement sending tool call results to the OpenAI server based on reading context updates. This lets us use the normal assistant context aggregator and not a special OpenAI Realtime subclass that pushes up a special frame for function call results.
2025-10-29 15:36:58 -04:00
Paul Kompfner
47756319be
Update OpenAIRealtimeLLMService to work with LLMContext and LLMContextAggregatorPair (cont'd).
...
Receiving a new context (via a context frame) no longer serves as a signal to reset the conversation. That’s because we’re now receiving new contexts from the user aggregator every time new messages are added, and from the assistant aggregator when function call results come in. The code pattern we're heading towards, of “diffing” each new context with the previous on, sets us up for doing more sophisticated things in the future, like sending specific messages to OpenAI to edit its internally-tracked context.
Also, remove code that was directly modifying context.
2025-10-29 15:36:58 -04:00
Paul Kompfner
5fa56df014
Update OpenAIRealtimeLLMService to work with LLMContext and LLMContextAggregatorPair (cont'd).
...
Update 19b example with new pattern.
2025-10-29 15:36:58 -04:00
Paul Kompfner
8a151235c3
Update OpenAIRealtimeLLMService to work with LLMContext and LLMContextAggregatorPair (cont'd).
...
Deprecate `send_transcription_frames`—transcription frames are now always sent.
2025-10-29 15:36:57 -04:00
Paul Kompfner
ec42f8c24e
Update OpenAIRealtimeLLMService to work with LLMContext and LLMContextAggregatorPair (cont'd).
...
Push `TranscriptionFrame`s upstream, to be handled by the user context aggregator. This will require at least a couple of other changes:
- Updating examples to put transcript processors upstream from `OpenAIRealtimeLLMService`
- Maybe figuring out a way to preserve backward compatibility with existing pipelines that put transcript processors downstream from `OpenAIRealtimeLLMService`
- Updating `OpenAIRealtimeLLMService` to ignore new received context frames, since the upstream user context aggregator will generate those after each newly-added user message; hopefully nobody was reliant on the old behavior of resetting the conversation upon receiving a new context!
2025-10-29 15:36:57 -04:00
Paul Kompfner
29fd17b9ff
Update OpenAIRealtimeLLMService to work with LLMContext and LLMContextAggregatorPair (cont'd).
...
Avoid pushing `LLMTextFrame` when `OpenAIRealtimeLLMService` is configured to output audio. This avoids duplicate text in assistant messages in context. Conceptually, a speech-to-speech service encapsulates TTS behavior; in a "traditional" pipeline, `LLMTextFrames` are swallowed by the TTS service, so they should similarly not be pushed by a speech-to-speech service. Only. `TTSTextFrame`s should be pushed.
2025-10-29 15:36:57 -04:00
Paul Kompfner
3ea1e357f2
Update OpenAIRealtimeLLMService to work with LLMContext and LLMContextAggregatorPair (initial part of work)
2025-10-29 15:36:57 -04:00
kompfner
351ef617ae
Merge pull request #2932 from pipecat-ai/pk/gemini-live-universal-llmcontext
...
Update `GeminiLLMService` to work with `LLMContext` and `LLMContextAg…
2025-10-29 15:35:13 -04:00
Paul Kompfner
9dafb715c4
Update some deprecation versions
2025-10-29 15:30:43 -04:00
Paul Kompfner
82d494d3d4
Fix a bug in GeminiLiveLLMService related to ending gracefully—i.e. waiting for the bot to stop responding before ending the pipeline—when the service is configured with the TEXT modality
2025-10-29 14:34:02 -04:00
Mark Backman
e893aaa620
Merge pull request #2931 from ivaaan/hume-bugfix
...
Hume: use Octave v1 if description provided
2025-10-29 13:40:21 -04:00
Paul Kompfner
65c17a698e
Whoops - fix a bug in GeminiLiveLLMService where we weren't checking if a tool call result was already handled before reporting it to the LLM
2025-10-29 12:44:00 -04:00
Paul Kompfner
615aae5b95
Fix GeminiLiveLLMService's sending of LLMFullResponseStartFrame and LLMFullResponseEndFrame so that they properly bookend responses.
...
Properly bookended responses now work with:
- AUDIO modality (validated with 26b example)
- TEXT modality (validated with 26d example)
- AUDIO modality with Vertex AI (validated with 26h example)
It doesn't seem that TEXT modality is supported with Vertex AI, hence the missing "quadrant" of validation.
2025-10-29 12:33:37 -04:00
Ivan A
2f1061f300
Merge branch 'main' into hume-bugfix
2025-10-29 17:06:50 +01:00
ivaaan
9307079af2
upd changelog
2025-10-29 17:05:41 +01:00
Mark Backman
efa64642a4
Merge pull request #2930 from pipecat-ai/mb/simli-constructor-update
...
Update Simli to align with Pipecat constructor norms
2025-10-29 11:50:11 -04:00
Mark Backman
ede6c32149
Update Simli to align with Pipecat constructor norms
2025-10-29 11:47:23 -04:00
Aleix Conchillo Flaqué
493d6bf91e
Merge pull request #2936 from pipecat-ai/aleix/daily-python-0.21.0
...
pyproject: update daily-python to 0.21.0
2025-10-28 18:25:25 -07:00
Aleix Conchillo Flaqué
aaebcae2e8
pyproject: update daily-python to 0.21.0
2025-10-28 17:23:37 -07:00
Mark Backman
4d82a1260b
Merge pull request #2933 from pipecat-ai/mb/remove-aiohttp-session-sarvam
...
Remove aiohttp_session arg from SarvamTTSService
2025-10-28 16:54:56 -04:00
Paul Kompfner
f974c66e12
Update GeminiLLMService to work with LLMContext and LLMContextAggregatorPair
2025-10-28 15:46:28 -04:00
Mark Backman
533372ed37
Remove aiohttp_session arg from SarvamTTSService
2025-10-28 15:39:14 -04:00
ivaaan
a9118eb2cd
use Octave 1 if description provided
2025-10-28 20:36:34 +01:00
Aleix Conchillo Flaqué
84ed2468e5
Merge pull request #2924 from pipecat-ai/aleix/daily-transport-remove-join-timeout
...
DailyTransport: don't timeout prematurely on join
2025-10-28 10:43:28 -07:00
Aleix Conchillo Flaqué
d82d855c20
DailyTransport: don't timeout prematurely on leave
2025-10-28 10:41:19 -07:00
Mark Backman
412ff2a4a1
Merge pull request #2929 from pipecat-ai/mb/cartesia-sonic-3
...
Update Cartesia's default model to sonic-3
2025-10-28 13:07:28 -04:00
Mark Backman
82ccc160fb
Merge pull request #2923 from pipecat-ai/mb/runner-no-proxy-required
...
Remove development runner requirement for proxy
2025-10-28 11:59:38 -04:00
Mark Backman
9ef60bd468
Update Cartesia's default model to sonic-3
2025-10-28 11:49:54 -04:00
Aleix Conchillo Flaqué
f3c4bf08dd
DailyTransport: don't timeout prematurely on join
2025-10-27 17:52:19 -07:00
Mark Backman
f2cfbee3c3
Remove development runner requirement for proxy
2025-10-27 16:18:31 -04:00
Vanessa Pyne
8b063116ab
Merge pull request #2921 from pipecat-ai/vp-azure-ex-cleanup
...
cleanup logger message
2025-10-27 12:59:08 -05:00
vipyne
8096e62b34
cleanup logger message
2025-10-27 11:27:30 -05:00
kompfner
20f4b0e8ff
Merge pull request #2914 from pipecat-ai/pk/gemini-function-calling-fixes
...
Gemini function calling fixes
2025-10-27 09:45:29 -04:00
Paul Kompfner
6feaf91789
Fix a bug in GeminiLLMAdapter's handling of Gemini-specific context messages
2025-10-27 09:42:24 -04:00
Mark Backman
91d3ae07b3
Merge pull request #2915 from Rickaym/fix--rounding-the-edges-of-observer-function-method-deprecation
...
fix: use correct property names
2025-10-24 19:42:34 -04:00
Pyae Sone Myo
71841f71ef
fix: use correct property names
2025-10-25 00:47:46 +06:30
