Qwen3-ASR

Qwen3-ASR is a state-of-the-art multilingual automatic speech recognition model. It runs on-device using Metal GPU acceleration via MLX, with 4-bit quantization for efficient memory usage. Available in 0.6B and 1.7B parameter variants.

Pipeline

The Qwen3-ASR inference pipeline processes audio through four stages:

Performance

M2 Max, 64 GB. RTF < 1.0 = faster than real-time.

Model Variants

CLI Usage

Transcribe an audio file with the default Qwen3-ASR model:

.build/release/audio transcribe recording.wav

Options

# Use the larger 1.7B model
.build/release/audio transcribe recording.wav --model 1.7b

# Specify language
.build/release/audio transcribe recording.wav --language en

# Streaming mode with partial results
.build/release/audio transcribe recording.wav --stream --partial

Swift API

Use the Qwen3ASR module to transcribe audio programmatically:

import Qwen3ASR

// Load the model (downloads from HuggingFace on first use)
let model = try await Qwen3ASRModel.loadFromHub()

// Transcribe an audio file
let result = try await model.transcribe(audioFile: "recording.wav")
print(result.text)

CoreML Encoder (Neural Engine)

Run the audio encoder on the Neural Engine via CoreML, with the text decoder on GPU via MLX. This hybrid approach lowers power consumption and frees the GPU for concurrent workloads.

import Qwen3ASR

let encoder = try await CoreMLASREncoder.fromPretrained()
let model = try await Qwen3ASRModel.fromPretrained()
let text = try model.transcribe(
    audio: samples, sampleRate: 16000,
    coremlEncoder: encoder
)

# CLI
.build/release/audio transcribe recording.wav --engine qwen3-coreml

INT8 palettized (180 MB, cosine similarity > 0.999) is the default. An INT4 variant (90 MB) is also available for size-constrained deployments.

Streaming Mode

Streaming mode uses VAD (voice activity detection) to segment audio into chunks and transcribe them incrementally. This is useful for long recordings or real-time processing.

# Stream with default segment size
.build/release/audio transcribe recording.wav --stream

# Control maximum segment duration
.build/release/audio transcribe recording.wav --stream --max-segment 15

# Show partial (in-progress) results as they arrive
.build/release/audio transcribe recording.wav --stream --partial

The --max-segment flag controls the maximum chunk duration in seconds. The --partial flag enables partial result output, showing words as they are decoded.

Supported Formats

Qwen3-ASR accepts the following audio formats. All input is automatically resampled to 16 kHz mono internally.

• WAV — uncompressed PCM
• M4A — AAC-encoded audio
• MP3 — MPEG Layer III
• CAF — Apple Core Audio Format