Replace block_size by block_duration in audio source constructors

src/diart/blocks/base.py

@@ -58,9 +58,6 @@ def get_file_padding(self, filepath: FilePath) -> Tuple[float, float]:
         left = utils.get_padding_left(file_duration + right, self.duration)
         return left, right
 
-    def optimal_block_size(self) -> int:
-        return int(np.rint(self.step * self.sample_rate))
-
 
 class Pipeline:
     @staticmethod

src/diart/console/client.py

@@ -15,7 +15,7 @@ def send_audio(ws: WebSocket, source: Text, step: float, sample_rate: int):
     # Create audio source
     source_components = source.split(":")
     if source_components[0] != "microphone":
-        audio_source = src.FileAudioSource(source, sample_rate)
+        audio_source = src.FileAudioSource(source, sample_rate, block_duration=step)
     else:
         device = int(source_components[1]) if len(source_components) > 1 else None
         audio_source = src.MicrophoneAudioSource(step, device)

src/diart/console/stream.py

@@ -40,13 +40,12 @@ def run():
     pipeline = pipeline_class(config)
 
     # Manage audio source
-    block_size = config.optimal_block_size()
     source_components = args.source.split(":")
     if source_components[0] != "microphone":
         args.source = Path(args.source).expanduser()
         args.output = args.source.parent if args.output is None else Path(args.output)
         padding = config.get_file_padding(args.source)
-        audio_source = src.FileAudioSource(args.source, config.sample_rate, padding, block_size)
+        audio_source = src.FileAudioSource(args.source, config.sample_rate, padding, config.step)
         pipeline.set_timestamp_shift(-padding[0])
     else:
         args.output = Path("~/").expanduser() if args.output is None else Path(args.output)

src/diart/inference.py

@@ -320,7 +320,7 @@ def run_single(
             filepath,
             pipeline.config.sample_rate,
             padding,
-            pipeline.config.optimal_block_size(),
+            pipeline.config.step,
         )
         pipeline.set_timestamp_shift(-padding[0])
         inference = StreamingInference(

src/diart/sources.py

@@ -55,23 +55,23 @@ class FileAudioSource(AudioSource):
     padding: (float, float)
         Left and right padding to add to the file (in seconds).
         Defaults to (0, 0).
-    block_size: int
-        Number of samples per chunk emitted.
-        Defaults to 1000.
+    block_duration: int
+        Duration of each emitted chunk in seconds.
+        Defaults to 0.5 seconds.
     """
     def __init__(
         self,
         file: FilePath,
         sample_rate: int,
         padding: Tuple[float, float] = (0, 0),
-        block_size: int = 1000,
+        block_duration: float = 0.5,
     ):
         super().__init__(Path(file).stem, sample_rate)
         self.loader = AudioLoader(self.sample_rate, mono=True)
         self._duration = self.loader.get_duration(file)
         self.file = file
         self.resolution = 1 / self.sample_rate
-        self.block_size = block_size
+        self.block_size = int(np.rint(block_duration * self.sample_rate))
         self.padding_start, self.padding_end = padding
         self.is_closed = False
 
@@ -132,7 +132,7 @@ class MicrophoneAudioSource(AudioSource):
     Parameters
     ----------
     block_duration: int
-        Duration of each chunk emitted in seconds.
+        Duration of each emitted chunk in seconds.
         Defaults to 0.5 seconds.
     device: int | str | (int, str) | None
         Device identifier compatible for the sounddevice stream.
@@ -268,10 +268,10 @@ def __init__(
         sample_rate: int,
         streamer: StreamReader,
         stream_index: Optional[int] = None,
-        block_size: int = 1000,
+        block_duration: float = 0.5,
     ):
         super().__init__(uri, sample_rate)
-        self.block_size = block_size
+        self.block_size = int(np.rint(block_duration * self.sample_rate))
         self._streamer = streamer
         self._streamer.add_basic_audio_stream(
             frames_per_chunk=self.block_size,