feat(regen): Speak V2 TTS websocket, agent update-listen, Flux EOT tuning

.fern/metadata.json

@@ -16,8 +16,8 @@
       "skip_validation": true
     }
   },
-  "originGitCommit": "8dd6f48a06eee8c3985894f68ba3b554e5564d21",
+  "originGitCommit": "335af96251466afae7a9f71badb65c630a48626e",
   "originGitCommitIsDirty": true,
   "invokedBy": "manual",
-  "sdkVersion": "7.3.2"
+  "sdkVersion": "7.4.1"
 }

.fernignore

@@ -26,6 +26,7 @@ src/deepgram/agent/v1/socket_client.py
 src/deepgram/listen/v1/socket_client.py
 src/deepgram/listen/v2/socket_client.py
 src/deepgram/speak/v1/socket_client.py
+src/deepgram/speak/v2/socket_client.py
 
 # Backward-compat patch: AgentV1SettingsAgentContext schema restructure as of
 # 2026-05-05. The new schema nests messages under .context.messages; this file
@@ -125,13 +126,16 @@ src/deepgram/core/query_encoder.py
 
 # Hand-written custom tests
 tests/custom/test_agent_history.py
+tests/custom/test_agent_update_listen.py
 tests/custom/test_compat_aliases.py
+tests/custom/test_eot_thresholds_feature.py
 tests/custom/test_language_hint_compat.py
 tests/custom/test_language_hints_feature.py
 tests/custom/test_listen_v2_regen_constraints.py
 tests/custom/test_query_encoder.py
 tests/custom/test_secure_logging.py
 tests/custom/test_socket_client_shims.py
+tests/custom/test_speak_v2_socket.py
 tests/custom/test_text_builder.py
 tests/custom/test_transport.py
 tests/typecheck/compat_aliases.py

AGENTS.md

@@ -53,6 +53,7 @@ How to identify:
 
 Current temporarily frozen files:
 - `src/deepgram/speak/v1/socket_client.py` — optional message param defaults, broad exception catch
+- `src/deepgram/speak/v2/socket_client.py` — same (optional `send_flush`/`send_close` defaults, broad exception catch); new websocket TTS client added in the 2026-07-08 regen
 - `src/deepgram/listen/v1/socket_client.py` — same
 - `src/deepgram/listen/v2/socket_client.py` — same + `send_configure` typing.Any/raw shim, response Union uses typing.Any instead of `ListenV2ConfigureSuccess`
 - `src/deepgram/agent/v1/socket_client.py` — same + `_sanitize_numeric_types`

examples/25-text-to-speech-streaming-v2.py

@@ -0,0 +1,66 @@
+"""
+Example: Text-to-Speech Streaming with the Speak V2 (Flux) WebSocket
+
+This example shows how to stream text-to-speech conversion using the Speak V2
+WebSocket. Speak V2 uses Flux voices (model strings of the form
+`flux-{voice}-{language}`, e.g. `flux-alexis-en`); use Speak V1 for Aura voices.
+"""
+
+from typing import Union
+
+from dotenv import load_dotenv
+
+load_dotenv()
+
+from deepgram import DeepgramClient
+from deepgram.core.events import EventType
+from deepgram.speak.v2.types import SpeakV2Speak
+
+SpeakV2SocketClientResponse = Union[str, bytes]
+
+client = DeepgramClient()
+
+try:
+    with client.speak.v2.connect(model="flux-alexis-en", encoding="linear16", sample_rate="24000") as connection:
+
+        def on_message(message: SpeakV2SocketClientResponse) -> None:
+            if isinstance(message, bytes):
+                print("Received audio data")
+                # In production, you would write this audio data to a file or play it
+                # with open("output.raw", "ab") as audio_file:
+                #     audio_file.write(message)
+            else:
+                msg_type = getattr(message, "type", "Unknown")
+                print(f"Received {msg_type} event")
+
+        connection.on(EventType.OPEN, lambda _: print("Connection opened"))
+        connection.on(EventType.MESSAGE, on_message)
+        connection.on(EventType.CLOSE, lambda _: print("Connection closed"))
+        connection.on(EventType.ERROR, lambda error: print(f"Error: {error}"))
+
+        # For sync version: Send messages before starting to listen
+        # Note: start_listening() blocks, so send all messages first
+        # For better control with bidirectional communication, use the async version
+        connection.send_speak(SpeakV2Speak(text="Hello, this is a text to speech example."))
+
+        # Flush to ensure all text is processed
+        connection.send_flush()
+
+        # Close the connection when done
+        connection.send_close()
+
+        # Start listening - this blocks until the connection closes
+        # All messages should be sent before calling this in sync mode
+        connection.start_listening()
+
+    # For async version:
+    # from deepgram import AsyncDeepgramClient
+    # async with client.speak.v2.connect(...) as connection:
+    #     listen_task = asyncio.create_task(connection.start_listening())
+    #     await connection.send_speak(SpeakV2Speak(text="..."))
+    #     await connection.send_flush()
+    #     await connection.send_close()
+    #     await listen_task
+
+except Exception as e:
+    print(f"Error: {e}")

examples/README.md

@@ -25,6 +25,7 @@ This directory contains comprehensive examples demonstrating how to use the Deep
 - **22-text-builder-demo.py** - TextBuilder demo (no API key required)
 - **23-text-builder-helper.py** - TextBuilder with REST API generation
 - **24-text-builder-streaming.py** - TextBuilder with streaming TTS (WebSocket)
+- **25-text-to-speech-streaming-v2.py** - Streaming TTS via the Speak V2 (Flux) WebSocket
 
 ### 30-39: Voice Agent