Add support for Audio Events

speechmatics/cli.py

@@ -26,6 +26,7 @@
 from speechmatics.exceptions import JobNotFoundException, TranscriptionError
 from speechmatics.helpers import _process_status_errors
 from speechmatics.models import (
+    AudioEventsConfig,
     AudioSettings,
     AutoChaptersConfig,
     BatchLanguageIdentificationConfig,
@@ -341,6 +342,10 @@ def get_transcription_config(
     if args_auto_chapters or auto_chapters_config is not None:
         config["auto_chapters_config"] = AutoChaptersConfig()
 
+    arg_audio_events = args.get("audio_events")
+    if arg_audio_events:
+        config["audio_events_config"] = AudioEventsConfig()
+
     if args["mode"] == "rt":
         # pylint: disable=unexpected-keyword-arg
         return TranscriptionConfig(**config)
@@ -448,6 +453,14 @@ def transcript_handler(message):
             sys.stdout.write(f"{escape_seq}{plaintext}\n")
         transcripts.text += plaintext
 
+    def audio_event_handler(message):
+        if print_json:
+            print(json.dumps(message))
+            return
+        event_name = message["event"].get("type", "").upper()
+        sys.stdout.write(f"{escape_seq}[{event_name}]\n")
+        transcripts.text += f"[{event_name}] "
+
     def partial_translation_handler(message):
         if print_json:
             print(json.dumps(message))
@@ -480,6 +493,8 @@ def end_of_transcript_handler(_):
     # print both transcription and translation messages (if json was requested)
     # print translation (if text was requested then)
     # print transcription (if text was requested without translation)
+
+    api.add_event_handler(ServerMessageType.AudioEventStarted, audio_event_handler)
     if print_json:
         if enable_partials or enable_translation_partials:
             api.add_event_handler(

speechmatics/cli_parser.py

@@ -466,6 +466,12 @@ def get_arg_parser():
         choices=["none", "speaker", "speaker_change"],
         help="Which type of diarization to use.",
     )
+
+    rt_transcribe_command_parser.add_argument(
+        "--audio-events",
+        action="store_true",
+        help="Print audio events to the console"
+    )
 
     # Build our actual parsers.
     mode_subparsers = parser.add_subparsers(title="Mode", dest="mode")

speechmatics/client.py

@@ -72,7 +72,6 @@ def __init__(
             self.connection_settings.set_missing_values_from_config(UsageMode.RealTime)
         self.websocket = None
         self.transcription_config = None
-        self.translation_config = None
 
         self.event_handlers = {x: [] for x in ServerMessageType}
         self.middlewares = {x: [] for x in ClientMessageType}
@@ -135,12 +134,15 @@ def _set_recognition_config(self):
         :py:attr:`speechmatics.models.ClientMessageType.SetRecognitionConfig`
         message.
         """
+        assert self.transcription_config is not None
         msg = {
             "message": ClientMessageType.SetRecognitionConfig,
             "transcription_config": self.transcription_config.as_config(),
         }
-        if self.translation_config is not None:
-            msg["translation_config"] = self.translation_config.asdict()
+        if self.transcription_config.translation_config is not None:
+            msg["translation_config"] = self.transcription_config.translation_config.asdict()
+        if self.transcription_config.audio_events_config is not None:
+            msg["audio_events_config"] = self.transcription_config.audio_events_config.asdict()
         self._call_middleware(ClientMessageType.SetRecognitionConfig, msg, False)
         return msg
 
@@ -155,13 +157,16 @@ def _start_recognition(self, audio_settings):
         :param audio_settings: Audio settings to use.
         :type audio_settings: speechmatics.models.AudioSettings
         """
+        assert self.transcription_config is not None
         msg = {
             "message": ClientMessageType.StartRecognition,
             "audio_format": audio_settings.asdict(),
             "transcription_config": self.transcription_config.as_config(),
         }
-        if self.translation_config is not None:
-            msg["translation_config"] = self.translation_config.asdict()
+        if self.transcription_config.translation_config is not None:
+            msg["translation_config"] = self.transcription_config.translation_config.asdict()
+        if self.transcription_config.audio_events_config is not None:
+            msg["audio_events_config"] = self.transcription_config.audio_events_config.asdict()
         self.session_running = True
         self._call_middleware(ClientMessageType.StartRecognition, msg, False)
         LOGGER.debug(msg)
@@ -435,7 +440,6 @@ async def run(
             consumer/producer tasks.
         """
         self.transcription_config = transcription_config
-        self.translation_config = transcription_config.translation_config
         self.seq_no = 0
         self._language_pack_info = None
         await self._init_synchronization_primitives()

speechmatics/models.py

@@ -211,6 +211,10 @@ class TopicDetectionConfig:
 class AutoChaptersConfig:
     """Auto Chapters config."""
 
+@dataclass
+class AudioEventsConfig:
+    def asdict(self):
+        return asdict(self)
 
 @dataclass(init=False)
 class TranscriptionConfig(_TranscriptionConfig):
@@ -254,12 +258,16 @@ class TranscriptionConfig(_TranscriptionConfig):
     """Indicates if partial translation, where words are produced
     immediately, is enabled."""
 
-    translation_config: TranslationConfig = None
+    translation_config: Optional[TranslationConfig] = None
     """Optional configuration for translation."""
+
+    audio_events_config: Optional[AudioEventsConfig] = None
+    """Optional configuration for audio events"""
 
     def as_config(self):
         dictionary = self.asdict()
         dictionary.pop("translation_config", None)
+        dictionary.pop("audio_events_config", None)
         dictionary.pop("enable_translation_partials", None)
         enable_transcription_partials = dictionary.pop(
             "enable_transcription_partials", False
@@ -504,6 +512,13 @@ class ServerMessageType(str, Enum):
     AddTranscript = "AddTranscript"
     """Indicates the final transcript of a part of the audio."""
 
+    AudioEventStarted = "AudioEventStarted"
+    """Indicates the start of an audio event."""
+
+    AudioEventEnded = "AudioEventEnded"
+    """Indicates the end of an audio event."""
+
+
     AddPartialTranslation = "AddPartialTranslation"
     """Indicates a partial translation, which is an incomplete translation that
     is immediately produced and may change as more context becomes available.