codeclawd/parakeet-tdt-0.6b-v3-coreml

Mirror of FluidInference/parakeet-tdt-0.6b-v3-coreml at revision 7dd20fe6b1797d35f5e3307e8b1732d9a178edfe, kept by the PrivateWispr team as a backup host. Licence: CC-BY-4.0 (as declared in this card's metadata; the original body text below says Apache 2.0, but the base model nvidia/parakeet-tdt-0.6b-v3 is CC-BY-4.0). Original authors: NVIDIA (parakeet-tdt-0.6b-v3 model, CC-BY-4.0) and FluidInference (Core ML conversion). All model files are byte-identical to the upstream revision; only this header was added to the README.

🧃 parakeet-tdt-0.6b-v3: Multilingual Speech-to-Text Model CoreML

| | | |

On‑device multilingual ASR model converted to Core ML for Apple platforms. This model powers FluidAudio’s batch ASR and is the same model used in our backend. It supports 25 European languages and is optimized for low‑latency, private, offline transcription.

The conversion script is here: https://github.com/FluidInference/mobius/tree/main/models/stt/parakeet-tdt-v3-0.6b/coreml

And the benchmarks are here: https://github.com/FluidInference/FluidAudio/blob/main/Documentation/Benchmarks.md#transcription

Highlights

• Core ML: Runs fully on‑device (ANE/CPU) on Apple Silicon.
• Multilingual: 25 European languages; see model usage in FluidAudio for examples.
• Performance: ~110× RTF on M4 Pro for batch ASR (1 min audio ≈ 0.5 s).
• Privacy: No network calls required once models are downloaded.

Intended Use

• Batch transcription of complete audio files on macOS/iOS.
• Local dictation and note‑taking apps where privacy and latency matter.
• Embedded ASR in production apps via the FluidAudio Swift framework.

Supported Platforms

• macOS 14+ (Apple Silicon recommended)
• iOS 17+

Model Details

• Architecture: Parakeet TDT v3 (Token Duration Transducer, 0.6B parameters)
• Input audio: 16 kHz, mono, Float32 PCM in range [-1, 1]
• Languages: 25 European languages (multilingual)
• Precision: Mixed precision optimized for Core ML execution (ANE/CPU)

Performance

• Real‑time factor (RTF): ~110× on M4 Pro in batch mode
• Throughput and latency vary with device, input duration, and compute units (ANE/CPU).

Usage

For quickest integration, use the FluidAudio Swift framework which handles model loading, audio preprocessing, and decoding.

Swift (FluidAudio)

import AVFoundation
import FluidAudio

Task {
    // Download and load ASR models (first run only)
    let models = try await AsrModels.downloadAndLoad()

    // Initialize ASR manager with default config
    let asr = AsrManager(config: .default)
    try await asr.initialize(models: models)

    // Load audio and transcribe
    let samples = try await AudioProcessor.loadAudioFile(path: "path/to/audio.wav")
    let result = try await asr.transcribe(samples, source: .system)
    print(result.text)

    asr.cleanup()
}

For more examples (including CLI usage and benchmarking), see the FluidAudio repository: https://github.com/FluidInference/FluidAudio

Files

• Core ML model artifacts suitable for use via the FluidAudio APIs (preferred) or directly with Core ML.
• Tokenizer and configuration assets are included/managed by FluidAudio’s loaders.

Limitations

• Primary coverage is European languages; performance may degrade for non‑European languages.

License

Apache 2.0. See the FluidAudio repository for details and usage guidance.