cs.CLNov 8, 2026
SaveTID System for Speech-COSER Challenge 2026: ASR Track
Organizations: Telefónica Innovación Digital, Spain · Universidad Autónoma de Madrid, Spain
Abstract
We describe the system submitted by Telefónica Innovación Digital to the Speech-COSER Challenge 2026, which features spontaneous speech from elderly speakers recorded in rural Spain. For ASR, we perform CTC-based pseudo-forced alignment scores audio-text match quality to guide a two-stage LoRA curriculum, followed by post-hoc magnitude pruning of the LoRA matrices to enhance generalization. An ensemble of seven systems combined via ROVER achieves 36.25% unnormalized and 22.17% normalized WER.
Figures & tables
Figure 1: CTC alignment score distributions for pre-segmented clips and long-file turns. The vertical line marks the threshold separating low- and high-confidence utterances.
Figure 2: Post-hoc LoRA magnitude pruning on the internal test set. Curves show two stages WER across keep ratios .
| Internal | Official | |||
|---|---|---|---|---|
| System | WER | WER norm | WER | WER norm |
| Zero-shot | 49.20 | 35.63 | – | – |
| Initial LoRA submission | 47.63 | 33.34 | 45.47 | 31.61 |
| Best single system | 40.09 | 26.22 | 38.22 | 24.06 |
Table 1: Single-system results for Voxtral-Mini-3B across internal (720 segments) and official evaluation sets.
| Individual system (internal) | Ensemble after adding this voter | |||||||||||
| Zero-shot | Fine-tuned | Internal | Official | |||||||||
| # | Voter | Checkpoint | WER | WER norm | WER | WER norm | WER | WER norm | WER | WER norm | ||
| 1 | Voxtral-Mini-3B , | S2, | 49.20 | 35.63 | 40.09 | 26.22 | 40.09 | 26.22 | 38.22 | – | 24.06 | – |
| 2 | Voxtral-Mini-3B , | S2, | 49.20 | 35.63 | 40.52 | 26.62 | – | – | – | – | – | – |
| 3 | Qwen3-ASR-1.7B | S2, | 58.54 | 39.37 | 43.38 | 28.60 | 39.01 | 25.29 | – | – | – | – |
| 4 | omniASR-LLM-7B-v2 | S2, unpruned | 59.05 | 37.32 | 50.28 | 34.10 | 38.99 | 25.03 | 37.32 | 23.07 | ||
Table 2: Incremental ROVER ensemble. Voters are accumulated top to bottom. Individual system : internal test performance zero-shot and fine-tuned (S1/S2: curriculum stage; : LoRA keep ratio). Ensemble : cumulative pool evaluated internally and on official submissions; denotes official improvement over the preceding submission (row 4 relative to single-system row 1). All values are %.
More in this topic
Automatic Speech Recognition
Automatic speech recognition converts spoken language into text across speakers, languages, domains, and recording conditions.
See all 507 papersSpeech LLM post-training increasingly relies on efficient cross-modal alignment and robust low-resource adaptation, yet collecting large-scale audio-text pairs remains costly. Text-only alignment methods such as TASU reduce this burden by simulating CTC posteriors from transcripts, but they provide limited control over uncertainty and error rate, making curriculum design largely heuristic. We propose TASU2, a controllable CTC simulation framework that simulates CTC posterior distributions under a specified WER range, producing text-derived supervision that better matches the acoustic decoding interface. This enables principled post-training curricula that smoothly vary supervision difficulty without TTS. Across multiple source-to-target adaptation settings, TASU2 improves in-domain and out-of-domain recognition over TASU, and consistently outperforms strong baselines including text-only fine-tuning and TTS-based augmentation, while mitigating source-domain performance degradation.
The VoiceMOS Challenge 2026: Evaluating Speech Enhancement, Emotional TTS and Accented TTS Systems
We present the results of the VoiceMOS Challenge 2026, the fifth edition of a scientific challenge on automatic prediction of subjective speech assessments. After expanding the scope to music and general audio in 2025, we refocused the evaluation target on speech and organized three tracks: prediction of absolute and comparative category ratings for enhanced speech, prediction of naturalness and emotion-related tasks for emotional text-to-speech systems, and prediction of speaker and accent similarity for codec-based speech synthesis systems. The challenge attracted a total of 18 teams worldwide, with most teams successfully surpassing the provided baselines. We summarize the challenge results, representative top-performing systems, participant feedback, and directions for future editions.
ISCSLP 2026 CoT-TTS Challenge: Chain-of-Thought Reasoning for Context-Aware Text-to-Speech
Recent advances in text-to-speech (TTS) have greatly improved speech naturalness, speaker similarity, and controllability. However, most existing controllable TTS systems still rely on explicit user-provided style prompts, making it difficult to automatically determine how a sentence should be spoken in long and complex conversational scenarios. This proposal introduces the ISCSLP 2026 CoT-TTS Challenge, which aims to evaluate whether a system can infer the intended speaking manner from contextual information and generate speech consistent with both the reasoning output and the surrounding scene. The challenge contains two tracks: text-context-aware CoT-TTS and audio-context-aware CoT-TTS. We construct a large-scale bilingual training set from speech-rich media and provide carefully filtered evaluation data for leaderboard comparison. Each system is required to output both a chain-of-thought reasoning analysis and the generated speech waveform. The official evaluation combines objective metrics, multimodal LLM-based evaluation, and human subjective assessment. To facilitate reproducibility, we provide inference code together with a fine-tuning recipe for a 0.6B Qwen3-based model trained via a three-stage strategy. This challenge is expected to support research on context understanding, chain-of-thought reasoning, and expressive speech generation for applications such as film dubbing, audiobook production, virtual characters, and spoken dialogue agents. Further information about the associated challenge is available at:https://iscslp2026-cot-tts.github.io/challenge-website/