Speech-to-text Sourced statement

Whisper tiny on Raspberry Pi 5

Benchmark pending. We haven't measured this pair yet. Below is what the sources say about this hardware, quoted with links.

Hobbyist single-board computer. Common for Home Assistant voice and offline assistants. Small models only.

EvidenceSourced statement

Commercial useYes

Licenseapache-2.0

Stated memory~1 GB VRAM

Device memory8 GB RAM

Results

Benchmark pending. No measured numbers from us yet for Whisper tiny on Pi 5. Get notified or send us your numbers (with the method).

What the sources say about Pi 5

Third-party docs whisper.cpp lists Raspberry Pi among its supported platforms. source
Third-party docs Home Assistant: on a Raspberry Pi 4, Whisper takes around 8 seconds to process incoming voice commands (model size not stated). source
Caveat: Pi 4, not Pi 5; model size not stated

Method

We haven't run this pair. Everything on this page is quoted from the linked sources, with the source type labelled (model authors vs third parties). Vendor numbers are often measured under favourable conditions.

Ways to run Whisper tiny

Runtime	Precision	Documented devices
openai-whisper (PyTorch)	fp16/fp32	CPU, CUDA	source
faster-whisper (CTranslate2)	int8, fp16, fp32	CPU, CUDA	source
whisper.cpp (ggml)	fp16, quantized (q5/q8)	CPU (x86/ARM), Apple Metal + Core ML, CUDA, Vulkan, OpenVINO	source
Hugging Face transformers	fp16/fp32	CPU, CUDA	source

Run it

pip install pywhispercpp          # prebuilt whisper.cpp wheel (CPU)
ffmpeg -i input.mp3 -ar 16000 -ac 1 audio16k.wav   # whisper.cpp wants 16 kHz WAV
python -c "from pywhispercpp.model import Model; print(' '.join(s.text for s in Model('tiny', n_threads=4).transcribe('audio16k.wav')))"

We ran this exact command on our test box on 2026-10-21.

Caveats

Pi 4, not Pi 5; model size not stated