Local ASR model

Whistle

Q: Can Whistle run locally?

Whistle is listed by LocalClaw as a local ASR option. Hardware fit depends on runtime, model size and backend support.

Catalogue summary: Apache-licensed on-device ASR model for English, German, French, Spanish, Italian, Dutch and Polish. The official release ships a single 16.9 MB `.cact` weight file, live-stream transcription, word timestamps, speech embeddings and keyword biasing through the same Cactus Needle CPU/WebAssembly runtime used for Needle.

Repository editorial metadata; verify comparative claims in the linked upstream material.

Edge readyspeech-to-text transcription7 languagesApache 2.0

Choose an app

Start with the recommended app, or choose another supported option.

Compare speech models

Open model filesRecommended verified starting point Open in UnslothLaunches Unsloth Desktop on this model Open Python packageVerified package page without exposing a command Open on Hugging FaceFiles, licence and model card Check my machineOptional: verify the fit with your saved hardware

Desktop app links require the app to be installed. If nothing opens, LocalClaw will show app-download and model-file fallbacks.

Catalogue quality

9/10

Catalogue speed

9.8/10

Model size

16.9 MB .cact weights; no GPU required

Voices

N/A (streaming ASR: outputs text, word timestamps and embeddings)

Can Whistle run locally?

Whistle can run locally for offline speech-to-text. Use the verified setup options on this page; no terminal command is required to choose the right path.

Apache 2.0 license. Still verify upstream usage notes before shipping.

Upstream source

streamingrealtimemultilinguallow-latency

Audio profile

Cat. quality

Cat. speed

9.8

Audio

9.3

Best fit

Whistle is best for offline transcription, speech indexing and local voice pipelines.

Hardware: cpuappleedge

Model details

Type

Local ASR model

Family

cactus

Latency

ultra-low

Formats

cactwebassemblypython

Languages

en, de, fr, es, it, nl, pl

Context

16 kHz mono input, one-pass 30-second windows with longer windowed audio, live stream commit mode, language detection or forced language, keyword biasing and `needle.Whistle()` local object API; upstream does not publish a fixed RAM floor, but the model is a 16.9 MB CPU/edge file

Install locally

Check runtimeConfirm the backend supports cact, webassembly, python on your machine.

Open recommended setupUse the app and model links above. LocalClaw does not expose a terminal command.

Test locallyRun a short private audio prompt before moving into production workflows.

Good for

speech-to-text transcription
Edge ready local workflows
streaming, realtime, multilingual

Watch before shipping

Validate pronunciation, latency and artifacts with your own voice samples.
Review the upstream license and acceptable-use notes.
Benchmark on your target CPU, Apple Silicon or GPU setup.

Related TTS and speech models

CompareBrowse all TTS models Local AIBrowse LLM models macOS appGet LocalClaw