A text-based quality estimation model for speech translation, fine-tuned from COMETKiwi on IWSLT/WMT data. Takes a source text and hypothesis translation and returns a quality score.
Speech translation models are increasingly capable of preserving speech-specific information (e.g. speaker gender, prosody, and emphasis), yet evaluation metrics remain blind to such phenomena. We meta-evaluate both text- and speech-based quality estimation metrics on two contrastive datasets targeting gender agreement and prosody, and find that both fall short, even when given direct access to the speech signal. We then train SpeechCOMET, a family of quality estimation models with speech encoders, and evaluate a state-of-the-art SpeechLLM as a judge. Both match or exceed text-based COMET on standard quality estimation, but neither consistently assesses speech-specific phenomena. We identify three causes: (1) speech-specific features are not reliably preserved in current encoders, (2) models tend to ignore the source signal, and (3) quality estimation training data contains too few relevant examples.
1git clone https://github.com/MaikeZuefle/speechCOMET.git
2cd speechCOMET
3pip install -e .
Requires Python 3.12+ and CUDA.
1import speechcomet
2
3model = speechcomet.load_from_checkpoint(speechcomet.download_model("maikezu/COMETKiwi-RoBERTa-IWSLT"))
4
5sample = {"src": "I love cake.", "mt": "Ich liebe Kekse."}
6scores = model.predict(samples=[sample], gpus=1, batch_size=8).scores
1@misc{züfle2026needspeechevaluatespeech,
2 title={Why We Need Speech to Evaluate Speech Translation},
3 author={Maike Züfle and Danni Liu and Vilém Zouhar and Jan Niehues},
4 year={2026},
5 eprint={2605.28227},
6 archivePrefix={arXiv},
7 primaryClass={cs.CL},
8 url={https://arxiv.org/abs/2605.28227},
9}
