Skip to content
Audio

Whisper Small Multilingual CS

a model for mixed speech recognition

2 min readupdated

A fine-tuned version of the Whisper Small model by Jevvellabs, optimized for processing sentences that mix German, English, Spanish, and Polish (code-switching). The model maintains high accuracy in monolingual automatic speech recognition (ASR) and does not generate false silence hallucinations. It was trained on Whisper Small using LoRA adapters (small add-ons to the base model) on the decoder while freezing the audio encoder. Useful for developers of voice assistants, translators, and transcription systems for multilingual users.

Key features

  • A 41.4% reduction in word error rate (WER) on code-switched text compared to the base model.
  • Improved recognition of technical terms and specific names.
  • Preservation of accuracy on monolingual data without significant regressions.
  • Support for Hugging Face Transformers formats and a ready-to-use whisper.cpp GGML F16 artifact.

What you can do

  • Recognize audio recordings with frequent switching between German, English, Spanish, and Polish within a single dialogue.
  • Run transcription on CPUs and mobile devices using the ready GGML file via whisper.cpp.
  • Integrate the model into Python speech processing pipelines using the standard transformers library.

Example

import torch
from transformers import pipeline

asr = pipeline(
    "automatic-speech-recognition",
    model="Jevvellabs/whisper-small-cs-de-en-es-pl",
    dtype=torch.float16 if torch.cuda.is_available() else torch.float32,
    device=0 if torch.cuda.is_available() else -1,
)

result = asr("mixed_de_en_es_pl_audio.wav")
print(result["text"])
Read moreDetails
Benchmarks: WER on code-switching dropped from 31.62% to 18.52%
Term recognition accuracy increased to 77.08%
Hardware requirements: standard for the Whisper Small model, GGML F16 file available (465 MB)
License: MIT
Discuss Telegram 1 23

Will it run?

Runs in F16

Fits without quantisation.

  • F16weights 0.5 GB · ready files1.0 GB
  • FP8 / INT8recommendedweights 0.2 GB0.7 GB
  • 4-bitweights 0.1 GB0.6 GB

Tip: go 8-bit.FP8 / INT8 is within about a percent of full precision in quality, takes half the memory and runs faster on RTX 40/50 and H100. Full precision only pays off for fine-tuning.

What to get for this model

Massed ComputeOpen Massed Compute

Your own GPU: FP8 / INT8 — from RTX 2060 · F16 — from RTX 2060

Massed Compute hourly prices, November 2026. The discounted price applies with the coupon at checkout.

Estimate: weights plus working memory. Long videos, high resolutions and long contexts need more.

Run it on Massed Compute

The script downloads the model to the rented machine. How to run it is in the model card.

  1. 1. Rent a machine with a fitting GPU

    Open Massed Compute
  2. 2. Paste the script into the machine's terminal

    Connect over SSH (the Massed Compute panel shows the command) or open a terminal on the machine's desktop.

    HF_TOKEN / CIVITAI_TOKENThe script starts with token lines: paste yours between the quotes. The Hugging Face token is for gated models, the Civitai one for Civitai models.

    # Jevvellabs/whisper-small-cs-de-en-es-pl
    mkdir -p ~/models
    curl -LsSf https://astral.sh/uv/install.sh | sh
    source "$HOME/.local/bin/env"
    uv tool install "huggingface_hub[cli]"
    hf download Jevvellabs/whisper-small-cs-de-en-es-pl --local-dir ~/models/whisper-small-cs-de-en-es-pl
    # The model is in ~/models. How to run it is in the model card (link above).
  3. 3. Use it

    The model is in ~/models. How to run it is in the model card (link above).

Stop the machine in the panel when you are done: it bills by the minute while it runs.

How to speed it up

Picked for this model type. Choose your hardware above to see what fits you exactly.

  • FlashAttention 44.0.0b40from November 18, 2026new release

    fastest exact attention on B200

    Kernels for Blackwell (beta). Exact attention without quantisation — fine for everything, LLMs included.

    How to enable: pip install flash-attn-4; vLLM and SGLang pick it up themselves.

    Earlier generations (2)
    • FlashAttention 3 — The Hopper (H100/H200) version with FP8.
    • FlashAttention 2 — The standard for RTX 30/40/50 and A100: exact attention without the memory cost of long contexts.
  • torch.compile2.15.0from October 28, 2026new release

    10–30% for free

    Compiles the model for your GPU. Slower first run, faster afterwards.

    How to enable: model = torch.compile(model); in ComfyUI the TorchCompileModel node.

Model family

New posts about versions of whisper-small (quants, finetunes, LoRAs) will show up under the bell.

Quantised versions 12All on Hugging Face
Fine-tunes 12All on Hugging Face
LoRAs and adapters 6All on Hugging Face

About the model

Jevvellabs/whisper-small-cs-de-en-es-pl
Author
Jevvellabs
Category
Audio
Released
Posted

On NeyroLearn

Hugging Face

231

Ask AI about this model

Answered by Gemini 4 · the chat is kept only in this tab

Comments