{"slug": "fluidaudio", "name": "FluidAudio", "category": "ai", "category_name": "AI models and assistants", "description": "", "recommenders": 6, "commenters": 10, "warners": 0, "score": 6.361, "rising": -1.96, "first": "2026-02-27", "latest": "2026-10-21", "why": "FluidAudio is a big part of this and is actually used in two places during a session.", "why_is_quote": true, "has_verdict": false, "sentiment": {"recommend": 7, "neutral": 3, "compare": 1, "warn": 0}, "spark": [0, 0, 0, 1, 2, 2, 0, 2, 0, 0, 0, 0], "listed": true, "loved": [], "caveat": "", "short_quote": "FluidAudio is a big part of this and is actually used in two places during a session.", "kind": "", "icon": true, "domain": null, "github_repo": "github.com/fluidinference/fluidaudio", "official_url": "https://github.com/FluidInference/FluidAudio", "pricing": "", "aliases": ["FluidAudio"], "verdict": null, "praise": [{"label": "", "summary": "", "quotes": [{"id": 782460129, "item_id": 48903758, "url": "https://news.ycombinator.com/item?id=48903758", "by": "hamza_q_", "date": "2026-07-14", "stance": "recommend", "quote": "Contributing this pipelined setup to FluidAudio would be good.", "excerpt": "Investigated this and it turned out to be an amusing bug: audio decoding was happening three times instead of just once lol. I've put up a PR to remove the wasteful redundant decoding: https://github.com/FluidInference/FluidAudio/pull/799 With the updated PR code, ran a test comparing transcribing (using Parakeet V3) a 1 hr stereo 44.1 kHz mp3 vs the same audio in 16 kHz mono wav format. The result was about 21.3% sl", "context": "", "use_case": "speech transcription framework", "switched_from": "", "story": "Apple's new SpeechAnalyzer API, benchmarked against Whisper and its predecessor"}, {"id": 776574913, "item_id": 48535932, "url": "https://news.ycombinator.com/item?id=48535932", "by": "AG342", "date": "2026-06-14", "stance": "recommend", "quote": "FluidAudio is a big part of this and is actually used in two places during a session.", "excerpt": "Crash recovery is definitely something that I want to spend a bit more time on. I'm not entirely sure how Trace handles crashing right in the middle of a recording, so I'm going to put a bit of time aside in the next few days to properly explore this and see if I can come up with an elegant solution to it. I think I've got the other two bits covered. I pushed an update yesterday that adds active echo cancellation so ", "context": "", "use_case": "transcription engine", "switched_from": "", "story": null}, {"id": 776532624, "item_id": 48533289, "url": "https://news.ycombinator.com/item?id=48533289", "by": "jv22222", "date": "2026-06-14", "stance": "recommend", "quote": "I'm seeing a lot right here", "excerpt": "I'm seeing a lot right here: https://github.com/FluidInference/FluidAudio", "context": "", "use_case": "", "switched_from": "", "story": null}, {"id": 762704256, "item_id": 47669016, "url": "https://news.ycombinator.com/item?id=47669016", "by": "rahimnathwani", "date": "2026-04-06", "stance": "recommend", "quote": "There's a fork of FluidAudio that supports the recent Cohere model", "excerpt": "There's a fork of FluidAudio that supports the recent Cohere model: https://github.com/altic-dev/FluidAudio/tree/B/cohere-coreml-asr It's used by this dictation app: https://github.com/altic-dev/FluidVoice/", "context": "", "use_case": "", "switched_from": "", "story": "Show HN: Ghost Pepper \u2013 100% local hold-to-talk speech-to-text for macOS"}, {"id": 762689680, "item_id": 47668105, "url": "https://news.ycombinator.com/item?id=47668105", "by": "lloyd-christmas", "date": "2026-04-06", "stance": "recommend", "quote": "Or write your own custom one with the library that backs it", "excerpt": "Or write your own custom one with the library that backs it: https://github.com/FluidInference/FluidAudio I did that so that I could record my own inputs and finetune parakeet to make it accurate enough to skip post-processing.", "context": "", "use_case": "On-device speech library", "switched_from": "", "story": "Show HN: Ghost Pepper \u2013 100% local hold-to-talk speech-to-text for macOS"}]}], "caveats": [], "who": [], "tags": [], "use_cases": ["component of memoe", "running asr on apple devices", "speech transcription framework", "transcription engine", "on-device speech library", "audio ml"], "compared": [{"slug": "nvidia-parakeet", "name": "NVIDIA Parakeet", "co": 7, "switched_from": 0, "switched_to": 0}, {"slug": "speechanalyzer", "name": "SpeechAnalyzer", "co": 3, "switched_from": 0, "switched_to": 0}, {"slug": "macwhisper", "name": "MacWhisper", "co": 2, "switched_from": 0, "switched_to": 0}, {"slug": "coreml", "name": "CoreML", "co": 2, "switched_from": 0, "switched_to": 0}, {"slug": "mlx", "name": "MLX", "co": 2, "switched_from": 0, "switched_to": 0}], "makers": [{"id": 782406272, "item_id": 48900392, "url": "https://news.ycombinator.com/item?id=48900392", "by": "hamza_q_", "date": "2026-07-13", "stance": "recommend", "quote": "", "excerpt": "Recently contributed a patch to FluidAudio that sped up Parakeet V2 and V3 to 320x and 282x faster than real time, respectively: https://github.com/FluidInference/FluidAudio/pull/507 That means one hour of audio transcribed in 11.25 and 12.75 seconds. The Inscribe post doesn't give a speed factor for SpeechAnalyzer. However, this Argmax blog post reports 70: https://www.argmaxinc.com/blog/apple-and-argmax Based on th", "context": "", "use_case": "On-device speech recognition", "switched_from": "", "story": "Apple's new SpeechAnalyzer API, benchmarked against Whisper and its predecessor"}], "threads": [{"url": "https://news.ycombinator.com/item?id=48894752", "title": "Apple's new SpeechAnalyzer API, benchmarked against Whisper and its predecessor", "mentions": 4}, {"url": "https://news.ycombinator.com/item?id=48521236", "title": "item 48521236", "mentions": 3}, {"url": "https://news.ycombinator.com/item?id=47666024", "title": "Show HN: Ghost Pepper \u2013 100% local hold-to-talk speech-to-text for macOS", "mentions": 2}, {"url": "https://news.ycombinator.com/item?id=50187193", "title": "item 50187193", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=50011229", "title": "Ask HN: Got a Mac for running agents? what's your setup?", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=47326101", "title": "Show HN: RunAnwhere \u2013 Faster AI Inference on Apple Silicon", "mentions": 1}], "sentiment_by_month": {"2025-11": {}, "2025-12": {}, "2026-01": {}, "2026-02": {"recommend": 1}, "2026-03": {"recommend": 2}, "2026-04": {"recommend": 2}, "2026-05": {}, "2026-06": {"neutral": 1, "recommend": 2}, "2026-07": {"compare": 1}, "2026-08": {}, "2026-09": {}, "2026-10": {"neutral": 2}}, "quotes": [{"id": 802995460, "item_id": 50187216, "url": "https://news.ycombinator.com/item?id=50187216", "by": "yisraelgottz", "date": "2026-10-21", "stance": "neutral", "quote": "I took Hermes, FluidAudio, CUA Driver, WisprKit, and OpenWispr, put it all together", "excerpt": "I've been creating products for over a year now, and this was one of those products that changed my day-to-day life. I use Claude and ChatGPT all the time for work, and I was always sick and tired of going into a chat window to do what I needed to do. I wanted to just talk to Claude wherever I was in any app, so that's what I did. I took Hermes, FluidAudio, CUA Driver, WisprKit, and OpenWispr, put it all together, co", "context": "", "use_case": "component of Memoe", "switched_from": "", "story": null}, {"id": 800255601, "item_id": 50015975, "url": "https://news.ycombinator.com/item?id=50015975", "by": "aavisangle", "date": "2026-10-09", "stance": "neutral", "quote": "runs on-device through FluidAudio's AsrManager, pinned to CPU plus the Neural Engine.", "excerpt": "Parakeet TDT 0.6B v3 (int8 CoreML, about 470 MB) runs on-device through FluidAudio's AsrManager, pinned to CPU plus the Neural Engine.", "context": "", "use_case": "running ASR on Apple devices", "switched_from": "", "story": "Ask HN: Got a Mac for running agents? what's your setup?"}, {"id": 782460129, "item_id": 48903758, "url": "https://news.ycombinator.com/item?id=48903758", "by": "hamza_q_", "date": "2026-07-14", "stance": "recommend", "quote": "Contributing this pipelined setup to FluidAudio would be good.", "excerpt": "Investigated this and it turned out to be an amusing bug: audio decoding was happening three times instead of just once lol. I've put up a PR to remove the wasteful redundant decoding: https://github.com/FluidInference/FluidAudio/pull/799 With the updated PR code, ran a test comparing transcribing (using Parakeet V3) a 1 hr stereo 44.1 kHz mp3 vs the same audio in 16 kHz mono wav format. The result was about 21.3% sl", "context": "", "use_case": "speech transcription framework", "switched_from": "", "story": "Apple's new SpeechAnalyzer API, benchmarked against Whisper and its predecessor"}, {"id": 782357522, "item_id": 48897345, "url": "https://news.ycombinator.com/item?id=48897345", "by": "foobarqux", "date": "2026-07-13", "stance": "compare", "quote": "yap/Speechanalyzer is about 50% slower than fluidaudio on M1.", "excerpt": "Just tried test using yap on a single ~1hr mp3: yap/Speechanalyzer is about 50% slower than fluidaudio on M1. yap interface is nicer though. https://github.com/finnvoor/yap", "context": "", "use_case": "", "switched_from": "", "story": "Apple's new SpeechAnalyzer API, benchmarked against Whisper and its predecessor"}, {"id": 782352752, "item_id": 48897047, "url": "https://news.ycombinator.com/item?id=48897047", "by": "foobarqux", "date": "2026-07-13", "stance": "compare", "quote": "I'd like to know how SpeechAnalyzer compares in speed.", "excerpt": "Fluidaudio implements Parakeet on ANE. I'd like to know how SpeechAnalyzer compares in speed. https://github.com/FluidInference/FluidAudio", "context": "", "use_case": "", "switched_from": "", "story": "Apple's new SpeechAnalyzer API, benchmarked against Whisper and its predecessor"}, {"id": 776574913, "item_id": 48535932, "url": "https://news.ycombinator.com/item?id=48535932", "by": "AG342", "date": "2026-06-14", "stance": "recommend", "quote": "FluidAudio is a big part of this and is actually used in two places during a session.", "excerpt": "Crash recovery is definitely something that I want to spend a bit more time on. I'm not entirely sure how Trace handles crashing right in the middle of a recording, so I'm going to put a bit of time aside in the next few days to properly explore this and see if I can come up with an elegant solution to it. I think I've got the other two bits covered. I pushed an update yesterday that adds active echo cancellation so ", "context": "", "use_case": "transcription engine", "switched_from": "", "story": null}, {"id": 776532624, "item_id": 48533289, "url": "https://news.ycombinator.com/item?id=48533289", "by": "jv22222", "date": "2026-06-14", "stance": "recommend", "quote": "I'm seeing a lot right here", "excerpt": "I'm seeing a lot right here: https://github.com/FluidInference/FluidAudio", "context": "", "use_case": "", "switched_from": "", "story": null}, {"id": 776531521, "item_id": 48533220, "url": "https://news.ycombinator.com/item?id=48533220", "by": "blopker", "date": "2026-06-14", "stance": "neutral", "quote": "most of these apps are built around", "excerpt": "Nice! I really like how many variations on this idea are coming out. MacWhisper used to be great, but is kinda of a buggy mess now. I'm making my own, for personal use. I did a survey of many and they all (that I could find) skip the fundamentals. The major issues that I've run into: - Crash recovery. Most of these apps are incredibly buggy and crash all the time, taking the recorded audio with them. Macwhisper is in", "context": "", "use_case": "", "switched_from": "", "story": null}, {"id": 762704256, "item_id": 47669016, "url": "https://news.ycombinator.com/item?id=47669016", "by": "rahimnathwani", "date": "2026-04-06", "stance": "recommend", "quote": "There's a fork of FluidAudio that supports the recent Cohere model", "excerpt": "There's a fork of FluidAudio that supports the recent Cohere model: https://github.com/altic-dev/FluidAudio/tree/B/cohere-coreml-asr It's used by this dictation app: https://github.com/altic-dev/FluidVoice/", "context": "", "use_case": "", "switched_from": "", "story": "Show HN: Ghost Pepper \u2013 100% local hold-to-talk speech-to-text for macOS"}, {"id": 762689680, "item_id": 47668105, "url": "https://news.ycombinator.com/item?id=47668105", "by": "lloyd-christmas", "date": "2026-04-06", "stance": "recommend", "quote": "Or write your own custom one with the library that backs it", "excerpt": "Or write your own custom one with the library that backs it: https://github.com/FluidInference/FluidAudio I did that so that I could record my own inputs and finetune parakeet to make it accurate enough to skip post-processing.", "context": "", "use_case": "On-device speech library", "switched_from": "", "story": "Show HN: Ghost Pepper \u2013 100% local hold-to-talk speech-to-text for macOS"}, {"id": 757248036, "item_id": 47328002, "url": "https://news.ycombinator.com/item?id=47328002", "by": "pzo", "date": "2026-03-10", "stance": "recommend", "quote": "Other popular repos for such tasks I would recommend:", "excerpt": "FWIW this RCLI is only MIT license but their engine MetalRT is commercial. Not sure the license of their models I guess also not MIT. So IMHO this repo is misleading. Not sure why they decided to reinvent the wheel and write yet another ML engine (MetalRT) which is proprietary. I would most likely bet on CoreML since it have support for ANE (apple NPU) or MLX. Other popular repos for such tasks I would recommend: htt", "context": "", "use_case": "audio ML", "switched_from": "", "story": "Show HN: RunAnwhere \u2013 Faster AI Inference on Apple Silicon"}, {"id": 756162689, "item_id": 47260168, "url": "https://news.ycombinator.com/item?id=47260168", "by": "kermitime", "date": "2026-03-05", "stance": "recommend", "quote": "The devs are responsive and active and nice on their discord too.", "excerpt": "the parakeet TDT models that are coreml optimized by fluid audio are hands down the fastest local models i\u2019ve tried\u2014 worth checking out! (unloading to the NPU is where the edge is) https://huggingface.co/FluidInference/parakeet-tdt-0.6b-v2-coreml https://github.com/FluidInference/FluidAudio The devs are responsive and active and nice on their discord too. You\u2019ll find discussions on all the latest whizbangs with VAD, ", "context": "", "use_case": "CoreML speech models on Apple", "switched_from": "", "story": "Nvidia PersonaPlex 7B on Apple Silicon: Full-Duplex Speech-to-Speech in Swift"}, {"id": 754867249, "item_id": 47179203, "url": "https://news.ycombinator.com/item?id=47179203", "by": "pzo", "date": "2026-02-27", "stance": "recommend", "quote": "Can use this library", "excerpt": "You probably still better use inference on ANE (Apple Neural Engine) via CoreML rather than Metal - speed will be either similar or even faster on non-pro macbooks or iphones and power consumption significantly better. Metal or even MLX format doesn't have to be the fastest and the only way to access ANE is via CoreML. Can use this library: https://github.com/FluidInference/FluidAudio", "context": "", "use_case": "", "switched_from": "", "story": "Parakeet.cpp \u2013 Parakeet ASR inference in pure C++ with Metal GPU acceleration"}, {"id": 734919329, "item_id": 45932458, "url": "https://news.ycombinator.com/item?id=45932458", "by": "hamza_q_", "date": "2025-11-14", "stance": "recommend", "quote": "Fluid Inference also have great implementations of Parakeet v2/v3 in CoreML for Apple devices", "excerpt": "Cool use of ONNX! Fluid Inference also have great implementations of Parakeet v2/v3 in CoreML for Apple devices and OpenVINO for Intel: https://github.com/FluidInference/FluidAudio https://github.com/FluidInference/eddy-audio", "context": "", "use_case": "On-device speech recognition", "switched_from": "", "story": "Show HN: Chirp \u2013 Local Windows dictation with ParakeetV3 no executable required"}]}