https://twango.dev/writingtwango.dev2026-06-16T00:00:00.000Zfeed (SvelteKit)James Dingjames@twango.devhttps://twango.devWriting from twango.devAll rights reserved, James Ding<![CDATA[Jetplay: Audio and Video Playback for JetBrains IDEs]]>https://twango.dev/writing/jetplay2026-06-16T00:00:00.000ZA plugin to allow playback of audio and video inside JetBrains IDEs (IntelliJ, PyCharm, WebStorm, and the rest), right in the editor tab.
]]>James Dingjames@twango.devhttps://twango.dev2026-06-16T00:00:00.000Z<![CDATA[The TartanAviation ATC Collection: Audio, ADS-B, and 531k Labels]]>https://twango.dev/writing/tartanaviation-atc-labels2026-06-05T00:00:00.000ZThree Hugging Face datasets from CMU's TartanAviation air traffic control corpus: paired audio and ADS-B clips, a half-million-utterance VAD split, and a full ASR label set. Why decode-time biasing and LLM correction lost to a plain ROVER vote, and the one signal that moved label quality.
]]>James Dingjames@twango.devhttps://twango.dev2026-06-05T00:00:00.000Z<![CDATA[How Good is Gemma 4 at ASR Tasks? Benchmarking Against the Open ASR Leaderboard]]>https://twango.dev/writing/gemma4-asr-benchmark2026-06-04T00:00:00.000ZBenchmarking the Gemma 4 family (E2B, E4B, and the new 12B) across all eight Open ASR Leaderboard short-form datasets, batched. WER, RTFx, where a general-purpose multimodal LLM holds up, and where it doesn't.