feat: mid-stream language_codes updates, Opus streaming encodings, sync connection pre-warming (v0.64.25)

README.md

@@ -104,7 +104,7 @@ aai.settings.api_key = "YOUR_API_KEY"
 audio_file = "./example.mp3"
 
 config = aai.TranscriptionConfig(
-    speech_models=["universal-3-pro", "universal-2"],
+    speech_models=["universal-3-5-pro", "universal-2"],
     language_detection=True,
     speaker_labels=True,
 )
@@ -130,7 +130,7 @@ aai.settings.api_key = "YOUR_API_KEY"
 audio_file = "https://assembly.ai/wildfires.mp3"
 
 config = aai.TranscriptionConfig(
-    speech_models=["universal-3-pro", "universal-2"],
+    speech_models=["universal-3-5-pro", "universal-2"],
     language_detection=True,
     speaker_labels=True,
 )
@@ -177,7 +177,7 @@ aai.settings.api_key = "<YOUR_API_KEY>"
 audio_file = "https://assembly.ai/wildfires.mp3"
 
 config = aai.TranscriptionConfig(
-  speech_models=["universal-3-pro", "universal-2"],
+  speech_models=["universal-3-5-pro", "universal-2"],
   language_detection=True
 )
 
@@ -214,7 +214,7 @@ aai.settings.api_key = "<YOUR_API_KEY>"
 audio_file = "https://assembly.ai/wildfires.mp3"
 
 config = aai.TranscriptionConfig(
-  speech_models=["universal-3-pro", "universal-2"],
+  speech_models=["universal-3-5-pro", "universal-2"],
   language_detection=True
 )
 
@@ -248,7 +248,7 @@ aai.settings.api_key = "<YOUR_API_KEY>"
 audio_file = "https://assembly.ai/wildfires.mp3"
 
 config = aai.TranscriptionConfig(
-  speech_models=["universal-3-pro", "universal-2"],
+  speech_models=["universal-3-5-pro", "universal-2"],
   language_detection=True
 )
 
@@ -280,7 +280,7 @@ aai.settings.api_key = "<YOUR_API_KEY>"
 audio_file = "https://assembly.ai/wildfires.mp3"
 
 config = aai.TranscriptionConfig(
-  speech_models=["universal-3-pro", "universal-2"],
+  speech_models=["universal-3-5-pro", "universal-2"],
   language_detection=True
 )
 config.set_custom_spelling(
@@ -324,7 +324,7 @@ aai.settings.api_key = "<YOUR_API_KEY>"
 audio_file = "https://assembly.ai/wildfires.mp3"
 
 config = aai.TranscriptionConfig(
-  speech_models=["universal-3-pro", "universal-2"],
+  speech_models=["universal-3-5-pro", "universal-2"],
   language_detection=True
 )
 
@@ -402,7 +402,7 @@ aai.settings.api_key = "<YOUR_API_KEY>"
 audio_file = "https://assembly.ai/wildfires.mp3"
 
 config = aai.TranscriptionConfig(
-    speech_models=["universal-3-pro", "universal-2"],
+    speech_models=["universal-3-5-pro", "universal-2"],
     language_detection=True,
 ).set_redact_pii(
     policies=[
@@ -431,7 +431,7 @@ aai.settings.api_key = "<YOUR_API_KEY>"
 audio_file = "https://assembly.ai/wildfires.mp3"
 
 config = aai.TranscriptionConfig(
-    speech_models=["universal-3-pro", "universal-2"],
+    speech_models=["universal-3-5-pro", "universal-2"],
     language_detection=True,
 ).set_redact_pii(
     policies=[
@@ -465,7 +465,7 @@ aai.settings.api_key = "<YOUR_API_KEY>"
 audio_file = "https://assembly.ai/wildfires.mp3"
 
 config = aai.TranscriptionConfig(
-    speech_models=["universal-3-pro", "universal-2"],
+    speech_models=["universal-3-5-pro", "universal-2"],
     language_detection=True,
     auto_chapters=True
 )
@@ -493,7 +493,7 @@ aai.settings.api_key = "<YOUR_API_KEY>"
 audio_file = "https://assembly.ai/wildfires.mp3"
 
 config = aai.TranscriptionConfig(
-  speech_models=["universal-3-pro", "universal-2"],
+  speech_models=["universal-3-5-pro", "universal-2"],
   language_detection=True,
   summarization=True,
   summary_model=aai.SummarizationModel.informative,
@@ -531,7 +531,7 @@ aai.settings.api_key = "<YOUR_API_KEY>"
 audio_file = "https://assembly.ai/wildfires.mp3"
 
 config = aai.TranscriptionConfig(
-    speech_models=["universal-3-pro", "universal-2"],
+    speech_models=["universal-3-5-pro", "universal-2"],
     language_detection=True,
     content_safety=True
 )
@@ -583,7 +583,7 @@ aai.settings.api_key = "<YOUR_API_KEY>"
 audio_file = "https://assembly.ai/wildfires.mp3"
 
 config = aai.TranscriptionConfig(
-    speech_models=["universal-3-pro", "universal-2"],
+    speech_models=["universal-3-5-pro", "universal-2"],
     language_detection=True,
     sentiment_analysis=True
 )
@@ -626,7 +626,7 @@ aai.settings.api_key = "<YOUR_API_KEY>"
 audio_file = "https://assembly.ai/wildfires.mp3"
 
 config = aai.TranscriptionConfig(
-    speech_models=["universal-3-pro", "universal-2"],
+    speech_models=["universal-3-5-pro", "universal-2"],
     language_detection=True,
     entity_detection=True
 )
@@ -655,7 +655,7 @@ aai.settings.api_key = "<YOUR_API_KEY>"
 audio_file = "https://assembly.ai/wildfires.mp3"
 
 config = aai.TranscriptionConfig(
-    speech_models=["universal-3-pro", "universal-2"],
+    speech_models=["universal-3-5-pro", "universal-2"],
     language_detection=True,
     iab_categories=True
 )
@@ -690,7 +690,7 @@ aai.settings.api_key = "<YOUR_API_KEY>"
 audio_file = "https://assembly.ai/wildfires.mp3"
 
 config = aai.TranscriptionConfig(
-    speech_models=["universal-3-pro", "universal-2"],
+    speech_models=["universal-3-5-pro", "universal-2"],
     language_detection=True,
     auto_highlights=True
 )
@@ -745,7 +745,7 @@ client.on(StreamingEvents.Termination, on_terminated)
 client.on(StreamingEvents.Error, on_error)
 
 client.connect(StreamingParameters(
-    sample_rate=16000, speech_model="universal-3-5-pro", format_turns=True,
+    sample_rate=16000, speech_model="universal-3-5-pro",
 ))
 try:
     client.stream(aai.extras.stream_file(filepath="audio.wav", sample_rate=16000))
@@ -874,7 +874,7 @@ async def main():
     async with AsyncStreamingClient(StreamingClientOptions(api_key="<YOUR_API_KEY>")) as client:
         client.on(StreamingEvents.Turn, on_turn)
         await client.connect(StreamingParameters(
-            sample_rate=16000, speech_model="universal-3-5-pro", format_turns=True,
+            sample_rate=16000, speech_model="universal-3-5-pro",
         ))
         await client.stream(stream_file_async("audio.wav", 16000))
 
@@ -1032,7 +1032,7 @@ transcriber = aai.Transcriber(config=config)
 transcriber.transcribe(
     "https://example.com/audio.mp3",
     # overrides the above configuration on the `Transcriber` with the following
-    config=aai.TranscriptionConfig(speech_models=["universal-3-pro", "universal-2"], multichannel=True, disfluencies=True)
+    config=aai.TranscriptionConfig(speech_models=["universal-3-5-pro", "universal-2"], multichannel=True, disfluencies=True)
 )
 ```
 

assemblyai/__version__.py

@@ -1 +1 @@
-__version__ = "0.64.21"
+__version__ = "0.64.25"

assemblyai/client.py

@@ -47,10 +47,17 @@ def __init__(
         def _store_response(response):
             self._last_response = response
 
+        keepalive_expiry = self.settings.keepalive_expiry
+        limits = (
+            httpx.Limits(keepalive_expiry=keepalive_expiry)
+            if keepalive_expiry is not None
+            else httpx.Limits()
+        )
         self._http_client = httpx.Client(
             base_url=self.settings.base_url,
             headers=headers,
             timeout=self.settings.http_timeout,
+            limits=limits,
             event_hooks={"response": [_store_response]},
         )
 

assemblyai/streaming/v3/_base.py

@@ -135,6 +135,11 @@ def _emit_param_warnings(params: StreamingParameters) -> None:
             "`customer_support_audio_capture=True` will record session audio. "
             "Only enable this when explicitly coordinating with AssemblyAI support."
         )
+    if params.language_code is not None:
+        logger.warning(
+            "[Deprecation Warning] `language_code` is deprecated and will be removed in a future release. "
+            "Please use `language_codes` instead."
+        )
 
 
 def _build_uri(host: str, params: StreamingParameters) -> str:

assemblyai/streaming/v3/models.py

@@ -147,11 +147,18 @@ class StreamingSessionParameters(BaseModel):
     agent_context: Optional[str] = None
     interruption_delay: Optional[int] = None
     turn_left_pad_ms: Optional[int] = None
+    language_codes: Optional[List[str]] = None
 
 
 class Encoding(str, Enum):
     pcm_s16le = "pcm_s16le"
     pcm_mulaw = "pcm_mulaw"
+    # Raw Opus packets, one packet per binary WS message. `sample_rate` is
+    # optional and ignored — the Opus stream is self-describing.
+    opus = "opus"
+    # Ogg-encapsulated Opus byte stream (ffmpeg, gstreamer, opusenc, browser
+    # MediaRecorder output). `sample_rate` is optional and ignored.
+    ogg_opus = "ogg_opus"
 
     def __str__(self):
         return self.value
@@ -259,6 +266,8 @@ class StreamingParameters(StreamingSessionParameters):
     sample_rate: int
     encoding: Optional[Encoding] = None
     speech_model: Optional[SpeechModel] = None
+    # Deprecated: use language_codes instead (pass a single-element list, e.g.
+    # ["es"], for the same behavior). Still supported for backward compatibility.
     language_code: Optional[str] = None
     language_detection: Optional[bool] = None
     domain: Optional[StreamingDomain] = None

assemblyai/sync.py

@@ -2,9 +2,11 @@
 
 import concurrent.futures
 import os
-from typing import BinaryIO, Optional, Tuple, Union
+from typing import Any, BinaryIO, Optional, Tuple, Union
 from urllib.parse import urlparse
 
+import httpx
+
 from . import client as _client
 from . import sync_api, types
 
@@ -201,3 +203,52 @@ def transcribe_async(
             data=data,
             config=config,
         )
+
+    def warm(self) -> bool:
+        """
+        Opens the connection to the sync API ahead of time.
+
+        The sync API is a single request/response, so a `transcribe()` that
+        opens its connection on demand pays the full DNS + TCP + TLS handshake
+        on the critical path — one network round trip that, for a distant
+        client, can rival the transcription itself. Calling `warm()` as soon as
+        you know audio is coming — typically while the clip is still being
+        recorded — spends that setup concurrently: the next `transcribe()`
+        reuses the already-open connection.
+
+        The warmed connection is reused while it stays in the HTTP pool —
+        `settings.keepalive_expiry` seconds (httpx's 5s default unless raised).
+        Call `warm()` shortly before `transcribe()`, or raise
+        `keepalive_expiry` (e.g. to 120, the sync audio cap) so a single call
+        covers a whole in-progress recording. `warm()` is idempotent and cheap,
+        so calling it again to refresh the connection is fine.
+
+        Routing the same `config.model` as the eventual transcription ensures
+        the warmed connection lands on the right backend.
+
+        Returns:
+            True once the connection is open (any HTTP response — even a
+            non-200 health probe — means the socket is established); False if
+            the connection could not be opened (transport error).
+        """
+        settings = self._client.settings
+        url = settings.sync_base_url.rstrip("/") + sync_api.ENDPOINT_HEALTH
+        try:
+            self._client.http_client.get(
+                url,
+                headers={sync_api.MODEL_HEADER: self.config.model},
+                timeout=min(settings.sync_http_timeout, 10.0),
+            )
+        except httpx.HTTPError:
+            return False
+        return True
+
+    def close(self) -> None:
+        """Shuts down the worker-thread pool used by `transcribe_async`."""
+        self._executor.shutdown(wait=False)
+
+    def __enter__(self) -> "SyncTranscriber":
+        return self
+
+    def __exit__(self, *_exc: Any) -> None:
+        self.close()

assemblyai/sync_api.py

@@ -6,6 +6,7 @@
 from . import types
 
 ENDPOINT_TRANSCRIBE = "/transcribe"
+ENDPOINT_HEALTH = "/healthz"
 MODEL_HEADER = "X-AAI-Model"
 
 

assemblyai/types.py

@@ -126,6 +126,9 @@ class Settings(BaseSettings):
     http_timeout: float = 30.0
     "The default HTTP timeout for general requests"
 
+    keepalive_expiry: Optional[float] = None
+    "Seconds an idle connection is kept in the HTTP pool before being closed. None uses httpx's default (5s). Raise it (e.g. to 120) so a connection opened by `SyncTranscriber.warm()` stays reusable across an in-progress recording."
+
     base_url: str = "https://api.assemblyai.com"
     "The base URL for the AssemblyAI API"