videosdk-plugins-openai 0.0.20__py3-none-any.whl → 0.0.22__py3-none-any.whl

videosdk/plugins/openai/realtime_api.py

@@ -45,9 +45,9 @@ DEFAULT_INPUT_AUDIO_TRANSCRIPTION = InputAudioTranscription(
 DEFAULT_TOOL_CHOICE = "auto"
 
 OpenAIEventTypes = Literal[
-    "instructions_updated",
-    "tools_updated",
-    "text_response"
+    "user_speech_started",
+    "text_response",
+    "error"
 ]
 DEFAULT_VOICE = "alloy"
 DEFAULT_INPUT_AUDIO_FORMAT = "pcm16"
@@ -305,11 +305,12 @@ class OpenAIRealtime(RealtimeBaseModel[OpenAIEventTypes]):
                 await self._handle_text_done(data)
 
         except Exception as e:
-            self.emit_error(f"Error handling event {event_type}: {str(e)}")
+            self.emit("error", f"Error handling event {event_type}: {str(e)}")
 
     async def _handle_speech_started(self, data: dict) -> None:
         """Handle speech detection start"""
         if "audio" in self.config.modalities:
+            self.emit("user_speech_started", {"type": "done"})
             await self.interrupt()
             if self.audio_track:
                 self.audio_track.interrupt()

videosdk/plugins/openai/tts.py

@@ -4,6 +4,7 @@ from typing import Any, AsyncIterator, Literal, Optional, Union
 import httpx
 import os
 import openai
+import asyncio
 
 from videosdk.agents import TTS
 
@@ -81,6 +82,7 @@ class OpenAITTS(TTS):
                 self.emit("error", "Audio track or event loop not set")
                 return
 
+            audio_data = b""
             async with self._client.audio.speech.with_streaming_response.create(
                 model=self.model,
                 voice=voice_id or self.voice,
@@ -91,13 +93,32 @@ class OpenAITTS(TTS):
             ) as response:
                 async for chunk in response.iter_bytes():
                     if chunk:
-                        self.loop.create_task(self.audio_track.add_new_bytes(chunk))
+                        audio_data += chunk
+
+
+            if audio_data:
+                await self._stream_audio_chunks(audio_data)
 
         except openai.APIError as e:
             self.emit("error", str(e))
         except Exception as e:
             self.emit("error", f"TTS synthesis failed: {str(e)}")
 
+    async def _stream_audio_chunks(self, audio_bytes: bytes) -> None:
+        """Stream audio data in chunks for smooth playback"""
+        chunk_size = int(OPENAI_TTS_SAMPLE_RATE * OPENAI_TTS_CHANNELS * 2 * 20 / 1000) 
+        
+        for i in range(0, len(audio_bytes), chunk_size):
+            chunk = audio_bytes[i:i + chunk_size]
+            
+            if len(chunk) < chunk_size and len(chunk) > 0:
+                padding_needed = chunk_size - len(chunk)
+                chunk += b'\x00' * padding_needed
+            
+            if len(chunk) == chunk_size:
+                self.loop.create_task(self.audio_track.add_new_bytes(chunk))
+                await asyncio.sleep(0.001)
+
     async def aclose(self) -> None:
         """Cleanup resources"""
         await self._client.close()

videosdk/plugins/openai/version.py

@@ -1 +1 @@
-__version__ = "0.0.20"
+__version__ = "0.0.22"

{videosdk_plugins_openai-0.0.20.dist-info → videosdk_plugins_openai-0.0.22.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: videosdk-plugins-openai
-Version: 0.0.20
+Version: 0.0.22
 Summary: VideoSDK Agent Framework plugin for OpenAI services
 Author: videosdk
 License-Expression: Apache-2.0
@@ -13,7 +13,7 @@ Classifier: Topic :: Multimedia :: Video
 Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
 Requires-Python: >=3.11
 Requires-Dist: openai[realtime]>=1.68.2
-Requires-Dist: videosdk-agents>=0.0.20
+Requires-Dist: videosdk-agents>=0.0.22
 Description-Content-Type: text/markdown
 
 # VideoSDK OpenAI Plugin

videosdk_plugins_openai-0.0.22.dist-info/RECORD

@@ -0,0 +1,9 @@
+videosdk/plugins/openai/__init__.py,sha256=1jbc4HOYxkLeruM9RAqmZYSBdnr74gnPHmCNMKXEPrg,259
+videosdk/plugins/openai/llm.py,sha256=h6xuJmyjg6InL9tr5pKBGt_5bNMpJ4XqnO72OtmCJ0c,7122
+videosdk/plugins/openai/realtime_api.py,sha256=WSzDWHcCQC8QsKLDmA5mm_oSN8UIHYMplesNliV5eUc,22611
+videosdk/plugins/openai/stt.py,sha256=YZROX-BjTqtWiT6ouMZacLkMYbmao3emB-88ewN93jg,9492
+videosdk/plugins/openai/tts.py,sha256=o5ktMUzjPkj64L5qqRaKPTWq7Na56TshMnLfU-sK36k,4417
+videosdk/plugins/openai/version.py,sha256=NoiGDztYD4fsDDnfSPiSzRkknkNHhFUtKZj0mhQiTYM,22
+videosdk_plugins_openai-0.0.22.dist-info/METADATA,sha256=9BJRuTdobykpCbIf5Gwr33z074lZjp-tCjdgBn5GUqg,827
+videosdk_plugins_openai-0.0.22.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
+videosdk_plugins_openai-0.0.22.dist-info/RECORD,,

videosdk_plugins_openai-0.0.20.dist-info/RECORD

@@ -1,9 +0,0 @@
-videosdk/plugins/openai/__init__.py,sha256=1jbc4HOYxkLeruM9RAqmZYSBdnr74gnPHmCNMKXEPrg,259
-videosdk/plugins/openai/llm.py,sha256=h6xuJmyjg6InL9tr5pKBGt_5bNMpJ4XqnO72OtmCJ0c,7122
-videosdk/plugins/openai/realtime_api.py,sha256=B2RlEV_yK0R4K1dPTyhhPewoa9bzd43ytEfsLKaHUUQ,22554
-videosdk/plugins/openai/stt.py,sha256=YZROX-BjTqtWiT6ouMZacLkMYbmao3emB-88ewN93jg,9492
-videosdk/plugins/openai/tts.py,sha256=LDsYXuHBoN-8g1iYt7JV_vRWOJZvhUN8QZQj_q264rU,3635
-videosdk/plugins/openai/version.py,sha256=cw-wPso5400rXRCR6WsHwthEUW8-b_VMrztjcYwBGfQ,22
-videosdk_plugins_openai-0.0.20.dist-info/METADATA,sha256=Ja6dfVaBHdrxMgmE0Hughw1oSzdMnbNPSOUfBptoopQ,827
-videosdk_plugins_openai-0.0.20.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
-videosdk_plugins_openai-0.0.20.dist-info/RECORD,,