Lib.rs

Cargo Features

transcribe-rs has no features set by default.

[dependencies]
transcribe-rs = { version = "0.3.11", features = ["audio-features", "onnx", "whisper-cpp", "whisper-metal", "whisper-vulkan", "whisper-cuda", "whisperfile", "openai", "vad-silero", "ort-cuda", "ort-tensorrt", "ort-directml", "ort-rocm", "ort-coreml", "ort-webgpu", "ort-xnnpack", "ort-tracing", "ort-accel", "all"] }

audio-features onnx?

Shared audio feature extraction (mel spectrogram, CTC decode, etc.)

Enables ndarray and rustfft

Affects transcribe-rs::decode, transcribe-rs::features…

onnx all? ort-coreml? ort-cuda? ort-directml? ort-rocm? ort-tracing? ort-webgpu? ort-xnnpack? = audio-features

ONNX-based models (SenseVoice, GigaAM, Parakeet, Moonshine, Canary)

Enables base64, once_cell, ort, and regex

ort:

ONNX runtime

Affects transcribe-rs::onnx…

whisper-cpp all? whisper-cuda? whisper-metal? whisper-vulkan?

Whisper via whisper.cpp (CPU-only by default; add whisper-metal, whisper-vulkan, or whisper-cuda for GPU)

Enables raw-api of whisper-rs

Whisper (GPU features controlled by whisper-metal / whisper-vulkan / whisper-cuda feature flags)

Affects transcribe-rs::whisper_cpp…

whisper-metal = whisper-cpp

--- Whisper Accelerators ---

Enables metal of whisper-rs

whisper-vulkan = whisper-cpp

Enables vulkan of whisper-rs

whisper-cuda = whisper-cpp

Enables cuda of whisper-rs

whisperfile all?

Whisperfile server

Enables ureq

Whisperfile

Affects transcribe-rs::whisperfile…

openai all?

Remote engines

Enables async-openai ^0.29.3, async-trait, and tokio

tokio:

OpenAI

Affects transcribe-rs::remote…

vad-silero

Silero neural VAD (requires silero_vad_v4.onnx model file)

Enables ndarray and ort

ort-cuda ort-accel? ort-tensorrt? = onnx

--- ORT Accelerators ---
Note: ort-cuda pulls in the CUDA execution provider, which adds ~800 MB+
to the ORT binary and requires a CUDA toolkit / cuDNN installation at runtime.

Enables cuda of ort

ort-tensorrt ort-accel? = ort-cuda

Enables tensorrt of ort

ort-directml ort-accel? = onnx

Enables directml of ort

ort-rocm ort-accel? = onnx

Enables rocm of ort

ort-coreml ort-accel? = onnx

Enables coreml of ort

ort-webgpu ort-accel? = onnx

Enables webgpu of ort

ort-xnnpack ort-accel? = onnx

Enables xnnpack of ort

ort-tracing = onnx

Enables tracing of ort

ort-accel = ort-coreml, ort-cuda, ort-directml, ort-rocm, ort-tensorrt, ort-webgpu, ort-xnnpack

all = onnx, openai, whisper-cpp, whisperfile

Convenience