feat: added write data to bytes

src/soundevent/audio/io.py

@@ -2,7 +2,7 @@
 
 Currently only supports reading and writing of .wav files.
 """
-
+from io import BytesIO
 from pathlib import Path
 from typing import Dict, Optional, Tuple
 
@@ -213,3 +213,22 @@ def load_clip(
             "samplerate": recording.samplerate,
         },
     )
+
+
+def audio_to_bytes(
+    data: np.ndarray,
+    samplerate: int,
+    bit_depth: int = 16,
+) -> bytes:
+    """Convert audio data to bytes."""
+    buffer = BytesIO()
+    with sf.SoundFile(
+        buffer,
+        mode="w",
+        samplerate=samplerate,
+        channels=data.shape[1],
+        format="RAW",
+        subtype=f"PCM_{bit_depth}",
+    ) as fp:
+        fp.write(data)
+    return buffer.getvalue()

src/soundevent/audio/media_info.py

@@ -1,5 +1,6 @@
 """Functions for getting media information from WAV files."""
 import hashlib
+import struct
 from dataclasses import dataclass
 from typing import IO
 
@@ -192,3 +193,56 @@ def compute_md5_checksum(path: PathLike) -> str:
             md5.update(buffer)
             buffer = fp.read(BUFFER_SIZE)
     return md5.hexdigest()
+
+
+def generate_wav_header(
+    samplerate: int,
+    channels: int,
+    samples: int,
+    bit_depth: int = 16,
+) -> bytes:
+    """Generate the data of a WAV header.
+
+    This function generates the data of a WAV header according to the
+    given parameters. The WAV header is a 44-byte string that contains
+    information about the audio data, such as the sample rate, the
+    number of channels, and the number of samples. The WAV header
+    assumes that the audio data is PCM encoded.
+
+    Parameters
+    ----------
+    samplerate
+        Sample rate in Hz.
+    channels
+        Number of channels.
+    samples
+        Number of samples.
+    bit_depth
+        The number of bits per sample. By default, it is 16 bits.
+
+    Notes
+    -----
+    The structure of the WAV header is described in
+    (WAV PCM soundfile format)[http://soundfile.sapp.org/doc/WaveFormat/].
+    """
+
+    data_size = samples * channels * bit_depth // 8
+    byte_rate = samplerate * channels * bit_depth // 8
+    block_align = channels * bit_depth // 8
+
+    return struct.pack(
+        "<4si4s4sihhiihh4si",  # Format string
+        b"RIFF",  # RIFF chunk id
+        data_size + 36,  # Size of the entire file minus 8 bytes
+        b"WAVE",  # RIFF chunk id
+        b"fmt ",  # fmt chunk id
+        16,  # Size of the fmt chunk
+        1,  # Audio format (1 corresponds to PCM)
+        channels,  # Number of channels
+        samplerate,  # Sample rate in Hz
+        byte_rate,  # Byte rate
+        block_align,  # Block align
+        bit_depth,  # Number of bits per sample
+        b"data",  # data chunk id
+        data_size,  # Size of the data chunk
+    )

src/soundevent/io/loader.py

@@ -23,7 +23,7 @@ def load(
     format: Optional[str] = "aoef",
     type: Literal["recording_set"] = "recording_set",  # type: ignore
 ) -> data.RecordingSet:  # type: ignore
-    ...
+    ...  # pragma: no cover
 
 
 @overload
@@ -33,7 +33,7 @@ def load(
     format: Optional[str] = "aoef",
     type: Literal["dataset"] = "dataset",  # type: ignore
 ) -> data.Dataset:  # type: ignore
-    ...
+    ...  # pragma: no cover
 
 
 @overload
@@ -43,7 +43,7 @@ def load(
     format: Optional[str] = "aoef",
     type: Literal["annotation_set"] = "annotation_set",  # type: ignore
 ) -> data.AnnotationSet:  # type: ignore
-    ...
+    ...  # pragma: no cover
 
 
 @overload
@@ -53,7 +53,7 @@ def load(
     format: Optional[str] = "aoef",
     type: Literal["annotation_project"] = "annotation_project",  # type: ignore
 ) -> data.AnnotationProject:  # type: ignore
-    ...
+    ...  # pragma: no cover
 
 
 @overload
@@ -63,7 +63,7 @@ def load(
     format: Optional[str] = "aoef",
     type: Literal["prediction_set"] = "prediction_set",  # type: ignore
 ) -> data.PredictionSet:  # type: ignore
-    ...
+    ...  # pragma: no cover
 
 
 @overload
@@ -73,7 +73,7 @@ def load(
     format: Optional[str] = "aoef",
     type: Literal["model_run"] = "model_run",  # type: ignore
 ) -> data.ModelRun:  # type: ignore
-    ...
+    ...  # pragma: no cover
 
 
 @overload
@@ -83,7 +83,7 @@ def load(
     format: Optional[str] = "aoef",
     type: Literal["evaluation_set"] = "evaluation_set",  # type: ignore
 ) -> data.EvaluationSet:  # type: ignore
-    ...
+    ...  # pragma: no cover
 
 
 @overload
@@ -93,7 +93,7 @@ def load(
     format: Optional[str] = "aoef",
     type: Literal["evaluation"] = "evaluation",  # type: ignore
 ) -> data.Evaluation:  # type: ignore
-    ...
+    ...  # pragma: no cover
 
 
 def load(

tests/test_audio/conftest.py

@@ -0,0 +1,12 @@
+"""Common fixtures for audio tests."""
+from pathlib import Path
+
+import pytest
+
+SAMPLE_AUDIO = Path(__file__).parent / "24bitdepth.wav"
+
+
+@pytest.fixture
+def sample_24_bit_audio() -> Path:
+    """Return a Path object to a 24-bit WAV file."""
+    return SAMPLE_AUDIO

tests/test_audio/test_io.py

@@ -0,0 +1,10 @@
+from pathlib import Path
+
+from soundevent.audio.io import audio_to_bytes, load_audio
+
+
+def test_audio_to_bytes(sample_24_bit_audio: Path):
+    original_bytes = sample_24_bit_audio.read_bytes()[44:]  # ignore header
+    audio, sr = load_audio(sample_24_bit_audio)
+    result_bytes = audio_to_bytes(audio, sr, bit_depth=24)
+    assert len(result_bytes) == len(original_bytes)

tests/test_audio/test_media_info.py

@@ -0,0 +1,35 @@
+"""Test suite for soundevent.audio.media_info module."""
+from pathlib import Path
+
+from soundevent.audio.media_info import (
+    MediaInfo,
+    generate_wav_header,
+    get_media_info,
+)
+
+
+def test_can_read_media_info(sample_24_bit_audio: Path):
+    media_info = get_media_info(sample_24_bit_audio)
+    assert isinstance(media_info, MediaInfo)
+    assert media_info.duration_s == 19.4015625
+    assert media_info.bit_depth == 24
+    assert media_info.samplerate_hz == 96000
+    assert media_info.channels == 1
+    assert media_info.samples == 1862550
+    assert media_info.audio_format == 1
+
+
+def test_can_generate_wav_header():
+    """Example WAV header from http://soundfile.sapp.org/doc/WaveFormat/."""
+    header = bytes.fromhex(
+        "52 49 46 46 24 08 00 00 57 41 56 45 66 6d 74 20 10 00 "
+        "00 00 01 00 02 00 22 56 00 00 88 58 01 00 04 00 10 00 "
+        "64 61 74 61 00 08 00 00 00 00 00 00 24 17 1e f3 3c 13 "
+        "3c 14 16 f9 18 f9 34 e7 23 a6 3c f2 24 f2 11 ce 1a 0d"
+    )
+    channels = 2
+    samplerate = 22050
+    bit_depth = 16
+    samples = 2048 // (channels * bit_depth // 8)
+    generated = generate_wav_header(samplerate, channels, samples, bit_depth)
+    assert header[:44] == generated