Mark Backman
6f1310569c
Add ElevenLabs output format type
2024-10-24 17:03:45 -04:00
Mark Backman
33553b71d4
Merge pull request #653 from pipecat-ai/mb/align-tts-constructors
...
Align TTSService constructors
2024-10-24 13:52:43 -04:00
Mark Backman
be8ca505cd
Merge pull request #652 from pipecat-ai/khk/more-gemini
...
Gemini new context manager and rewrite to use google data structures internally
2024-10-24 13:47:38 -04:00
Mark Backman
e957cce422
Align TTSService constructors
2024-10-24 13:42:06 -04:00
Mark Backman
418a13a4ec
Merge pull request #650 from pipecat-ai/mb/assembly-fix
...
AssemblyAI: don't disconnect on language change
2024-10-24 11:26:56 -04:00
Mark Backman
fc445c0a1f
Merge pull request #649 from pipecat-ai/mb/open-ai-max-tokens
...
Add max_tokens and max_completion_tokens inputs for OpenAI
2024-10-24 11:26:44 -04:00
Mark Backman
f0c65468ed
AssemblyAI: don't disconnect on language change
2024-10-24 08:30:48 -04:00
Mark Backman
ce6a2bdcf7
Add max tokens inputs to OpenAI
2024-10-24 07:03:45 -04:00
Kwindla Hultman Kramer
e032b0b70a
gemini context aggregators
2024-10-23 18:44:09 -07:00
Mark Backman
e39f7e965b
Support function calling for Grok
2024-10-23 17:22:26 -04:00
Aleix Conchillo Flaqué
801e52c095
transports(daily): add update_subscriptions()
2024-10-22 15:02:55 -07:00
Aleix Conchillo Flaqué
8d75fcc9f0
use warnings package to report deprecated code
2024-10-22 10:02:21 -07:00
Aleix Conchillo Flaqué
951255def9
transport(daily): use "nova-2-general" for transcription
2024-10-22 09:40:03 -07:00
Mark Backman
ccc3691620
Fix Silero VAD import issue
2024-10-21 20:39:20 -04:00
Mark Backman
46927805bc
Update GladiaSTTService to use the Gladia V2 API
2024-10-21 07:10:38 -04:00
Mark Backman
1dbf4ff27d
Add AssemblyAI STT service
2024-10-20 08:57:57 -04:00
Paul Vilchez
5640bd9447
Fixing a config mismatch which caused usage stats to only report when enable_metrics was true.
2024-10-20 03:33:13 -04:00
Aleix Conchillo Flaqué
be4bdabdf4
examples: use OpenAILLMContext in all the examples
2024-10-19 18:24:00 -07:00
Aleix Conchillo Flaqué
3502509d3e
move SileroVAD processor to processors package
2024-10-18 23:28:29 -07:00
Kwindla Hultman Kramer
07712cdb16
gemini function calling and partial implementation of standard context stuff
2024-10-18 17:14:57 -07:00
DamienDeepgram
13f232bafc
Update default model
2024-10-18 15:33:50 -07:00
Aleix Conchillo Flaqué
12eb9437c1
services(stt): allow STT service to passthrough audio
2024-10-17 16:15:09 -07:00
Aleix Conchillo Flaqué
8108423742
transport(livekit): force specifying a vad analyzer
...
Don't default to SileroVADAnalyzer(). Also, resample to input sample rate.
2024-10-17 14:06:43 -07:00
Aleix Conchillo Flaqué
d3f4ac61b6
move utils.audio to audio.utils and add resample_audio()
2024-10-17 13:59:32 -07:00
Aleix Conchillo Flaqué
2a37b2459a
move vad package to audio.vad
2024-10-17 13:49:16 -07:00
Mark Backman
d1000f2fe4
Merge pull request #606 from pipecat-ai/mb/add-playht-options
...
PlayHT: Add websocket TTS service; rename existing service to PlayHTHttpTTSService, upgrade client, add input params
2024-10-17 16:46:59 -04:00
Mark Backman
da3810f1a2
Add websocket support for PlayHT
2024-10-17 15:41:33 -04:00
Aleix Conchillo Flaqué
eb21597d1a
Merge pull request #603 from pipecat-ai/aleix/silero-vad-processor-fixes
...
vad: add support for interruption to SileroVAD processor
2024-10-17 10:48:39 -07:00
Aleix Conchillo Flaqué
e3eea0c02f
vad: add support for interruption to SileroVAD processor
2024-10-17 10:48:25 -07:00
Mark Backman
45606e177c
Add input options to PlayHT, upgrade to latest PlayHT model
2024-10-17 11:56:12 -04:00
Nathan Straub
d4ec6827ce
services(livekit) - Stop Sending EndFrame when Participant Disconnects
...
How It Works Now:
A participant disconnecting triggers and EndFrame, invoking stop() on the input and output transports and causing the LiveKit room to disconnect.
Proposal:
Match the daily implementation, and just trigger the callbacks in the LiveKitTransport. Leave it up to the implementor to decide whether to send EndFrames when this happens.
2024-10-16 23:53:31 -07:00
Aleix Conchillo Flaqué
e31d1152db
Merge pull request #601 from pipecat-ai/aleix/openai-realtime-misc
...
services(openai): rename OpenAILLMServiceRealtimeBeta to OpenAIRealti…
2024-10-16 16:20:18 -07:00
Mark Backman
55f1ae2564
Adjust log levels for log messages
2024-10-16 17:30:47 -04:00
Kwindla Hultman Kramer
93c9e219ce
fix for message handling bug on initialization
2024-10-16 12:40:20 -07:00
Aleix Conchillo Flaqué
edd44cc181
services(openai): rename OpenAILLMServiceRealtimeBeta to OpenAIRealtimeBetaLLMService
2024-10-16 10:20:19 -07:00
Mark Backman
af5a7e9092
Move metrics from transport to rtvi
2024-10-16 11:33:56 -04:00
Mark Backman
8d9a7486d1
Merge pull request #598 from pipecat-ai/mb/add-daily-metrics-message-frame
...
Comply with RTVI format for sending metrics data via Daily transport
2024-10-16 10:14:44 -04:00
Mark Backman
00d0f9ae48
Comply with RTVI format for sending metrics data
2024-10-16 09:00:38 -04:00
Aleix Conchillo Flaqué
3910aeb4de
transports(daily): don't send messages if not joined
2024-10-15 17:51:01 -07:00
Aleix Conchillo Flaqué
713dcb7a4d
transports(daily): cancel messages task when canceling
2024-10-15 17:51:01 -07:00
Aleix Conchillo Flaqué
04da51c7d8
transport(base_output): push EndFrame downstream at the right time
2024-10-15 17:51:01 -07:00
Aleix Conchillo Flaqué
e52d18e42d
processors(audiobuffer): make functions public
2024-10-15 15:31:59 -07:00
Aleix Conchillo Flaqué
4a71eacac3
rtvi: reset bot transcription with interruptions
2024-10-15 14:58:21 -07:00
Aleix Conchillo Flaqué
f0d89e57ad
frames: some frames need to be SystemFrames
...
We want to process user and bot started/stopped speaking frames as fast as
possible. If we queue them they might be processed too late.
2024-10-15 14:37:56 -07:00
Mark Backman
bb00dbefbc
More edge case handling for text filtering
2024-10-15 14:08:27 -04:00
Aleix Conchillo Flaqué
7bbaf4dfe9
rtvi: merge TTS/TTSText and LLM/LLMText processors
2024-10-15 10:24:43 -07:00
Aleix Conchillo Flaqué
3a3bf3fe34
services(cartesia): schedule TTSStoppedFrame after text
2024-10-15 10:06:28 -07:00
Aleix Conchillo Flaqué
616aa54f75
ruff formatting
2024-10-15 10:06:28 -07:00
Aleix Conchillo Flaqué
164f06415c
servcies(cartesia): no need to send LLMFullResponseEndFrame
...
Interruptions are already handled by context aggregators.
2024-10-15 10:06:28 -07:00
Aleix Conchillo Flaqué
51bc4839d1
transport(base_output): simplify code
2024-10-15 10:06:28 -07:00