Subtitle-Aligned Fine-Tuning of Whisper for Swiss German ASR: Benchmark Contamination, Convention Mismatch, and an Honest Baseline at 25.6% WER (13.8% cWER)
We present a systematic study of fine-tuning OpenAI's Whisper large-v3 for Swiss German ASR, using 1,367 hours of broadcast speech paired with Standard German subtitles as weak supervision. Through 16 iterative training runs on an NVIDIA DGX Spark (Grace Blackwell, 128 GB unified memory, up to 1 PFLOP FP4), we compare LoRA and full fine-tuning of the 1.55B-parameter model, investigate hallucination root causes, and quantify the effect of data quality, subtitle alignment, and training strategy. O
Record details
Published: 29 May 2026
Source: arXiv
Category: Research
Topics: Safety & alignment · Finance, VC & PE
Retrieved: 14 July 2026
Related evidence
These records share source-supplied organisations, an exact publisher byline, automatic topics or regions. The reason is shown on every link; related does not mean supporting, agreeing with or verifying this record.
Anthropic will deploy 2 gigawatts of AMD GPUs for Claude in a deal worth up to $5 billion
The Decoder · 22 July 2026
Nvidia, SpaceX, Microsoft launch AI safety initiative as OpenAI cyberattack fallout continues
CNBC Technology · 27 July 2026
Nvidia invests in Ilya Sutskever's AI lab, shifting SSI away from Google chips
The Decoder · 28 July 2026
Nvidia halved its $250bn promise to OpenAI. Then it showed investors what it owns
The Next Web AI · 15 August 2026
Investor pressure forces Nvidia to shrink its OpenAI bet just as Anthropic's numbers defy bubble warnings
The Decoder · 15 August 2026
Robots Are Coming — but Not Everywhere
MIT Sloan Management Review AI · 23 July 2026
How to cite this record
ethics.ai (29 May 2026), “Subtitle-Aligned Fine-Tuning of Whisper for Swiss German ASR: Benchmark Contamination, Convention Mismatch, and an Honest Baseline at 25.6% WER (13.8% cWER),” evidence record 3428, https://ethics.ai/record/3428 (originally published by arXiv).
Use and limitations
This page is a stable index and citation surface for a source record. ethics.ai did not author the underlying report and has not independently verified every claim. Automatic topics may be imperfect. For consequential use, quote and cite the original publisher.