# Gemma 4 speech encoder evaluated on Saudi and Gulf Arabic without weight changes

> Researchers tested a frozen Gemma 4 audio encoder and a lightweight adapter version on Arabic call‑center audio, finding comparable error rates to larger commercial models.

Oossa · 2026-10-22 · https://oossa.com/en/gemma-4-speech-encoder-evaluated-on-saudi-and-gulf-arabic-without-weight-changes

A team led by Amr Abdelkhalik examined the 305‑million‑parameter audio encoder used in Gemma 4 and EmbeddingGemma 2. They kept the encoder weights fixed and trained either a 22.6 M‑parameter CTC reader (frozen system) or added 3.1 M adapter parameters plus a 41.5 M hybrid CTC/attention reader (adapted system). Compared to Cohere Transcribe Arabic (2.07 B parameters), the frozen system matched character error rates on most tests and ran 3.4 × faster on CPU. The adapted system achieved statistically similar word error rates on Saudi customer‑service calls (19.20 % vs 19.27 %) and beat a 0.6 B‑parameter recogniser on several conditions, while decoding 139 × faster on one GPU.

## The facts

- Encoder size: 305 M parameters
- Adapted system adds 3.1 M adapter parameters and a 41.5 M reader

## Why it matters

The results show that a frozen or lightly adapted Gemma 4 encoder can deliver near‑commercial Arabic speech‑recognition accuracy with far lower compute, useful for companies that need cost‑effective, self‑hosted solutions.

## Sources & references

1. [How Far Can the Gemma 4 Speech Encoder Go Without Changing Its Weights? Saudi and Gulf Arabic Speech Recognition with a Frozen and an Adapter-Augmented Encoder](https://arxiv.org/abs/2610.28673) – arXiv, 2026-10-22

Last updated: 2026-10-22