Commit Graph

8304 Commits

Author SHA1 Message Date
Aleix Conchillo Flaqué
352361bdd2 Update changelog skill to avoid line wrapping 2026-02-19 09:20:33 -08:00
Aleix Conchillo Flaqué
baa61468a1 Add changelog entries for PR #3719 2026-02-19 09:20:33 -08:00
Aleix Conchillo Flaqué
7501ba2e45 Undeprecate DailyUpdateRemoteParticipantsFrame
Remove the deprecation warning and __post_init__ override. Also fix the
default value for remote_participants to use field(default_factory=dict)
instead of None.
2026-02-19 09:20:33 -08:00
Aleix Conchillo Flaqué
200716e8fe Add SIP transfer and SIP REFER frames to Daily transport
Add write_transport_frame() hook to BaseOutputTransport so subclasses
can handle custom frame types that flow through the audio queue. Add
DailySIPTransferFrame and DailySIPReferFrame as DataFrame subclasses
that queue with audio, ensuring SIP operations execute only after the
bot finishes its current utterance. Override write_transport_frame in
DailyOutputTransport to dispatch these frames to the existing
sip_call_transfer() and sip_refer() client methods.

Also switch DailyOutputTransport.send_message error handling from
logger.error to push_error for consistency.
2026-02-19 09:20:33 -08:00
Paul Kompfner
421696e1c2 Replace Any with specific types and add | _NotGiven to all *Settings field annotations across 49 service files
Every `*Settings` dataclass field whose default is `NOT_GIVEN` now carries `_NotGiven` in its type union so the type system accurately reflects the three-state semantics (real value, `None` where applicable, or not-yet-specified). Fields previously typed as bare `Any`, `str`, `float`, `bool`, `list`, `dict`, or `Optional[X]` are now narrowed to the specific type from the corresponding `InputParams` Pydantic model.
2026-02-19 11:28:29 -05:00
Mark Backman
50ef4909e3 Add changelog entries for PR #3774 2026-02-19 07:44:52 -07:00
Mark Backman
63df4642b5 Fix RTVIObserver missing upstream-only frames by adding broadcasted flag
RTVIObserver previously filtered out all upstream frames to avoid
duplicate messages from broadcasted frames. This caused upstream-only
frames to be silently ignored. Instead, add a `broadcasted` field to
the Frame base class that is set by broadcast_frame() and
broadcast_frame_instance(), and only skip upstream copies of
broadcasted frames.
2026-02-19 07:43:20 -07:00
Filipi da Silva Fuchter
43869a499d Merge pull request #3773 from pipecat-ai/mb/fix-ci-apt-get-update
Fix CI: add apt-get update before installing system packages
2026-02-19 09:28:25 -05:00
Mark Backman
d2bf3952ec Merge pull request #3772 from simliai/main
Update SimliClient to latest
2026-02-19 09:13:14 -05:00
Mark Backman
92c380ee77 Add apt-get update before installing system packages in CI
The CI was failing because the runner's package index was stale,
causing a 404 when fetching libasound2-dev (a dependency of
portaudio19-dev). Running apt-get update first refreshes the index.
2026-02-19 07:01:07 -07:00
antonyesk601
a55ba40921 fix: remove misimport 2026-02-19 10:41:17 +00:00
antonyesk601
fb1bfd03dd update SimliClient to latest 2026-02-19 10:35:50 +00:00
Paul Kompfner
a7edd8e441 Fix 55zp example 2026-02-18 17:15:22 -05:00
Paul Kompfner
2a07138abf Fix Grok Realtime dynamic session properties updating, and update corresponding 55zo example 2026-02-18 17:12:36 -05:00
Filipi da Silva Fuchter
a0a7b3101d Merge pull request #3765 from ianbbqzy/ian/inworld-default-async
[inworld] default timestamp transport strategy to ASYNC
2026-02-18 16:59:01 -05:00
Filipi da Silva Fuchter
39dc4ba99c Updated changelog/3765.changed.md 2026-02-18 16:58:27 -05:00
Paul Kompfner
ad942f6e4c Update 55zn example (UIltravox dynamic settings updates) to exercise changing modality, which is a setting that supports dynamic updates 2026-02-18 16:33:05 -05:00
Paul Kompfner
97d34ef9e1 Update OpenAI Realtime to warn when you try to update settings that can't be updated dynamically.
Update corresponding example to demonstrate updating output modality.
2026-02-18 16:16:06 -05:00
Paul Kompfner
c054780477 Fix 55zh example 2026-02-18 15:59:34 -05:00
Paul Kompfner
88a2dbdb82 Update 55zf example to update a setting that is supported by the default Camb TTS model 2026-02-18 15:48:50 -05:00
Paul Kompfner
d386a0efda Update Sarvam TTS to apply all changes to settings, not just voic 2026-02-18 15:31:08 -05:00
Paul Kompfner
b718a23c17 Tweak 55zd example 2026-02-18 15:25:50 -05:00
Paul Kompfner
e38f7d9451 Fix 55zc example 2026-02-18 15:23:23 -05:00
Paul Kompfner
b00d454842 Fix Inworld TTS settings updating 2026-02-18 15:19:57 -05:00
Paul Kompfner
0fa51811ea Fix 55z example 2026-02-18 15:11:04 -05:00
Paul Kompfner
323ee00b83 Fix 55w example 2026-02-18 14:51:48 -05:00
Paul Kompfner
0c73b77327 Update Lmnt TTS to support updating settings dynamically 2026-02-18 14:47:38 -05:00
Paul Kompfner
416e1cf877 Update Rime TTS services to store voice in the standard settings.voice field, as opposed to the nonstandard speaker field 2026-02-18 14:46:47 -05:00
Paul Kompfner
b4c5cb258b Tweak 55r example to make the settings update more pronounced 2026-02-18 14:15:14 -05:00
Paul Kompfner
728a97ade3 Update Deepgram TTS to support updating settings dynamically 2026-02-18 14:11:51 -05:00
Paul Kompfner
28677ec829 Tweak 55p example to make the settings update more pronounced 2026-02-18 13:49:32 -05:00
Paul Kompfner
17886d14e8 Fix ElevenLabsTTSService settings update code 2026-02-18 13:47:02 -05:00
Paul Kompfner
caf5dacbe8 Update 55j example to avoid console warning 2026-02-18 12:37:50 -05:00
Paul Kompfner
b8b531b66a In Cartesia TTS service, we don't need to override _update_settings. Parent class handling is enough, as new settings are picked up on the next run_tts (no need to reconnect). 2026-02-18 12:37:34 -05:00
Paul Kompfner
a14690e3a0 Fix the 55i example 2026-02-18 11:55:14 -05:00
Paul Kompfner
d913d954db Fix SpeechmaticsSTTService settings update code, and augment test file to better exercise it 2026-02-18 11:34:52 -05:00
Paul Kompfner
e98bb1df66 Simplify 55* examples: inline the settings update directly in the on_client_connected handler instead of wrapping it in a separate async task 2026-02-18 11:06:33 -05:00
Paul Kompfner
a7ada79fd9 Fix ElevenLabsRealtimeSTTService:
- Move `CommitStrategy` up in the file so it could be used by `ElevenLabsRealtimeSTTSettings`
- Fix a bug where `run_tts` would erroneously try to reconnect if a reconnection was already in flight (like a reconnection triggered by `_update_settings`)
2026-02-18 10:50:53 -05:00
Filipi da Silva Fuchter
a5b5a8e5cf Merge pull request #3759 from pipecat-ai/mb/gradium-context-update
Switch Gradium TTS to AudioContextWordTTSService for multiplexing
2026-02-18 10:16:57 -05:00
filipi87
1daea78b91 Fix GradiumTTSService to reuse context IDs across multiple run_tts calls and prevent the parent class from pushing text frames. 2026-02-18 12:12:49 -03:00
Paul Kompfner
7910f20e14 Update comment in Azure TTS explaining how we could support dynamic settings updates in the future 2026-02-18 10:07:33 -05:00
Paul Kompfner
d7d94a29f0 Add foundational examples (55) for runtime settings updates via *UpdateSettingsFrame
42 examples covering STT (13), TTS (21), LLM (4), and realtime (4) services. Each demonstrates updating service settings 10 seconds after client connects, verifying the typed settings machinery end-to-end for every provider.
2026-02-18 09:46:23 -05:00
Tanmay Chaudhari
6066eec853 Add changelog for PR #3768 2026-02-18 14:31:16 +05:30
Tanmay Chaudhari
cd379671aa fix: use audio.sample_rate instead of audio.audio_frames in TavusInputTransport 2026-02-18 14:18:16 +05:30
Ian Lee
8006223911 [inworld] default timestamp transport strategy to ASYNC 2026-02-17 15:13:20 -08:00
Paul Kompfner
ce51df677c Add backward-compat _aliases and from_mapping overrides to TTS settings
The migration from plain-dict `self._settings` to typed dataclasses renamed keys and flattened nested dicts. The deprecated dict-based `TTSUpdateSettingsFrame(settings={...})` code path calls `from_mapping`, which silently dropped old keys into `extra`.

- Add `_aliases` so renamed flat keys (e.g. `sample_rate` → `fish_sample_rate`, camelCase Inworld keys) resolve correctly.
- Override `from_mapping` to destructure nested dicts (`output_format`, `prosody`, `audioConfig`, `voice_setting`, `audio_setting`) into their flat field equivalents.
- Fix AsyncAI constructor bug passing `output_format={...}` dict instead of individual `output_container`/`output_encoding`/`output_sample_rate` fields.
2026-02-17 17:07:14 -05:00
Paul Kompfner
68ebd3d063 Migrate HumeTTSService to standard TTSSettings pattern and remove dead TTSService.update_setting
HumeTTSService now stores its params (description, speed, trailing_silence) in a proper `HumeTTSSettings` dataclass instead of a separate `_params` Pydantic model, making it work with `TTSUpdateSettingsFrame(update=...)`. The old `update_setting(key, value)` method is kept but deprecated.

Also removes the unused no-op `TTSService.update_setting` base method, which was never called by the `TTSUpdateSettingsFrame` pipeline.
2026-02-17 15:44:41 -05:00
Radek Sedlák
5ea2d47d39 feat: Add support for private endpoint in Azure STT 2026-02-17 21:42:00 +01:00
Paul Kompfner
94a651cee2 Remove dead ServiceSettings.to_dict method 2026-02-17 15:15:18 -05:00
Paul Kompfner
1cad4210ce Deprecate dict-based *UpdateSettingsFrame(settings={...}) code path in STT, TTS, and LLM services.
The dataclass-based API (`*UpdateSettingsFrame(update=*Settings(...))`) is the preferred path since 0.0.103. The dict path still works but now emits a `DeprecationWarning`.
2026-02-17 15:09:39 -05:00