wx44wx/pipecat
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Tweak changelog entries describing slate of recent updates to OpenAIRealtimeBetaLLMService

Browse Source
This commit is contained in:
Paul Kompfner
2025-03-18 15:52:11 -04:00
parent e707efbffa
commit 3dd4ef7230
1 changed files with 33 additions and 40 deletions
Download Patch File Download Diff File Expand all files Collapse all files

73
CHANGELOG.md
View File

@@ -128,51 +128,43 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
Gemini models. Added foundational example Gemini models. Added foundational example
`14p-function-calling-gemini-vertex-ai.py`. `14p-function-calling-gemini-vertex-ai.py`.
- Added support in `OpenAIRealtimeBetaLLMService` for the - Added support in `OpenAIRealtimeBetaLLMService` for a slate of new features:
`conversation.item.input_audio_transcription.delta` server message.
- Added error handling in `OpenAIRealtimeBetaLLMService` for the - The `'gpt-4o-transcribe-latest'` input audio transcription model.
`response.done` server message reporting a failure. - The `input_audio_noise_reduction` session property.
- Added support in `OpenAIRealtimeBetaLLMService` for the new ```python
`gpt-4o-transcribe-latest` input audio transcription model. session_properties = SessionProperties(
# ...
- Added support in `OpenAIRealtimeBetaLLMService` for the new input_audio_noise_reduction=InputAudioNoiseReduction(
`input_audio_noise_reduction` session property. type="near_field" # also supported: "far_field"
)
```python # ...
session_properties = SessionProperties(
# ...
input_audio_noise_reduction=InputAudioNoiseReduction(
type="near_field" # also supported: "far_field"
) )
# ... ```
)
```
- Added support in `OpenAIRealtimeBetaLLMService` for the new - The `'semantic_vad'` `turn_detection` session property value, a more
`semantic_vad` `turn_detection` session property, which is a more sophisticated model for detecting when the user has stopped speaking.
sophisticated model for detecting when the user has stopped speaking. - `on_conversation_item_created` and `on_conversation_item_updated`
events to `OpenAIRealtimeBetaLLMService`.
- Added `on_conversation_item_created` and `on_conversation_item_updated` ```python
events to `OpenAIRealtimeBetaLLMService`. @llm.event_handler("on_conversation_item_created")
async def on_conversation_item_created(llm, item_id, item):
# ...
```python @llm.event_handler("on_conversation_item_updated")
@llm.event_handler("on_conversation_item_created") async def on_conversation_item_updated(llm, item_id, item):
async def on_conversation_item_created(llm, item_id, item): # `item` may not always be available here
# ... # ...
```
@llm.event_handler("on_conversation_item_updated") - The `retrieve_conversation_item(item_id)` method for introspecting a
async def on_conversation_item_updated(llm, item_id, item): conversation item on the server.
# `item` may not always be available here
# ...
```
- Added `retrieve_conversation_item(item_id)` to `OpenAIRealtimeBetaLLMService`. ```python
item = await llm.retrieve_conversation_item(item_id)
```python ```
item = await llm.retrieve_conversation_item(item_id)
```
### Changed ### Changed
@@ -268,9 +260,10 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
- Fixed an issue in `RimeTTSService` where the last line of text sent didn't - Fixed an issue in `RimeTTSService` where the last line of text sent didn't
result in an audio output being generated. result in an audio output being generated.
- Fixed `OpenAIRealtimeBetaLLMService` by adding support for the - Fixed `OpenAIRealtimeBetaLLMService` by adding proper handling for:
`conversation.item.input_audio_transcription.delta` server message, which was - The `conversation.item.input_audio_transcription.delta` server message,
added server-side at some point and not handled client-side. which was added server-side at some point and not handled client-side.
- Errors reported by the `response.done` server message.
### Other ### Other