ffmpeg extension fails with unexpected options: threads for encoders that natively do not support threading.

[JacekDuszenko]

🐛 Describe the bug

Torchaudio ffmpeg cpp extension crashes when used with encoders that do not have threading capabilities e.g. mulaw. Easily reproducable by running the following script:

import io

from torchaudio.io import StreamReader

wave_i2_data = b"".join(
    [
        b"\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00",
        b"\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00",
    ]
)

io_bytes = io.BytesIO(wave_i2_data)
s = StreamReader(
    src=io_bytes,
    format="s16le",
    option={"sample_rate": str(8000)},
)
s.add_basic_audio_stream(
    frames_per_chunk=1024,
    sample_rate=8000,
    format="s16p",
    decoder="pcm_mulaw",
    decoder_option=None,
    num_channels=1,
)

Fails with:

  File "repro.py", line 18, in <module>
    s.add_basic_audio_stream(
  File ".venv/lib/python3.10/site-packages/torio/io/_streaming_media_decoder.py", line 681, in add_basic_audio_stream
    self.add_audio_stream(
  File ".venv/lib/python3.10/site-packages/torio/io/_streaming_media_decoder.py", line 778, in add_audio_stream
    self._be.add_audio_stream(
RuntimeError: Unexpected options: threads

The reason for that is the code sets the threads entry unconditionally in

audio/src/libtorio/ffmpeg/stream_reader/stream_processor.cpp

Line 139 in 318bace

if (!av_dict_get(opts, "threads", nullptr, 0)) {

while codecs without support for threading do not delete the entry afterwards. To fix the 'threads' keyword need to be deleted manually.

Versions

PyTorch version: N/A
Is debug build: N/A
CUDA used to build PyTorch: N/A
ROCM used to build PyTorch: N/A

OS: macOS 15.2 (arm64)
GCC version: Could not collect
Clang version: 16.0.0 (clang-1600.0.26.6)
CMake version: version 3.31.6
Libc version: N/A

Python version: 3.10.16 (main, Feb 12 2025, 15:03:04) [Clang 19.1.6 ] (64-bit runtime)
Python platform: macOS-15.2-arm64-arm-64bit
Is CUDA available: N/A
CUDA runtime version: Could not collect
CUDA_MODULE_LOADING set to: N/A
GPU models and configuration: Could not collect
Nvidia driver version: Could not collect
cuDNN version: Could not collect
HIP runtime version: N/A
MIOpen runtime version: N/A
Is XNNPACK available: N/A

CPU:
Apple M4 Max

Versions of relevant libraries:
[pip3] numpy==1.26.4
[conda] Could not collect