home.social

#aitranscription — Public Fediverse posts

Live and recent posts from across the Fediverse tagged #aitranscription, aggregated by home.social.

fetched live
  1. Longer-term take on Google’s Recorder app: a human in the loop remains needed

    Over the last six months, one of the apps I’ve used most on my phone has been one of the few that I can only use on my phone–Google’s Pixel-only Recorder. The free voice-transcription app that I latched onto after ditching Evernote has become a regular part of my workflow, even as it’s made its limits more obvious to me.

    Recorder’s core advantage remains the ability to work offline, which is good on both functional and privacy grounds. I’m stuck on horrible WiFi or in a cellular dead zone? No problem. A sensitive conversation with a source that I don’t want instantly uploaded to a cloud-based transcription service? Handled.

    (That second scenario has yet to actually come up in my work; if, however, somebody summons me to a clandestine meeting in a garage in Rosslyn, it’s nice to know I’ll be ready.)

    But maybe because this app has to rely on the processing power of the Pixel 9 Pro I bought nearly two years ago instead of borrowing any cloud smarts, it continues to make the same mistakes:

    • Its speaker recognition is more of a cloud of probability; in one transcript of a four-person panel last week, the app IDed me as “Speaker 1,”Speaker 2,” and “Speaker 3” while labeling only one quote as coming from “Speaker 4.”
    • It continue to capitalize random nouns–in particular, hitting its virtual Shift key for “Enterprise” when transcribing conversations about the computing needs of large organizations so often that I have to wonder if it was trained on Star Trek, German, or both.
    • Sometimes it just skips a sentence or two or three. I don’t know how that’s possible, considering there’s no break in the recorded audio, unless maybe my phone’s processor got briefly swamped.

    But even with these repeatable glitches, cleaning up a transcription rarely requires more than one playback while I have Recorder’s transcription open in an editing app–usually Google Docs, since that’s one of Recorder’s supported export options and since that Web app can itself work offline.

    Google’s updates to the app haven’t yet made a meaningful difference in its transcription accuracy but have added the option to get an AI-generated summary of a transcript. That, too, works offline, and I should try that more often.

    My overall take remains unchanged from what I wrote here in March: While this app could work better, I’d still want to/have to check any AI transcription against the original recording, so this one isn’t adding that much work. And it retains its non-trivial edge of costing nothing when I already pay enough for cloud services and can only expect those expenses to keep ratcheting up.

    #AI #AITranscription #GoogleRecorder #localAI #onDeviceAI #Pixel #Pixel9Pro #Recorder #workOffline
  2. Still chipping away at my note-taking migration from Evernote to Obsidian

    More than six months into the biggest switch of productivity apps I’ve undertaken since I stopped using Microsoft Word for everyday writing, I’m relieved to report that moving to a new note-taking app has so far involved much less drama than I expected. Or deserved, considering I staged this cutover right before my busiest workweek of the year.

    But my transition from Evernote to Obsidian–spurred by Evernote’s owners at the Italian firm Bending Spoons sticking me with a 92 percent rate increase that was going to land two days before the start of CES–isn’t complete either. Here’s how things have come along to date:

    • Import: I had a choice of note-taking apps that could import notes from the app I’d used regularly since 2010 (Microsoft’s OneNote not among them, ruling it out even though I already pay for it with my Microsoft 365 subscription). The biggest hassle with migrating some 15 years worth of notes turned out to be on Evernote’s end–its cap of 100 notes per export. Obsidian imported every batch without complaint but not without formatting glitches that I could not have avoided in an app based on the Markdown formatting system. And then I only realized Thursday morning that I had somehow failed to bring over about two years of notes from one notebook; about 10 minutes later, I had remedied that oversight.
    • Filename games: Obsidian’s implementation of the text format that Apple developer and blogger John Gruber released in 2004 (see Anil Dash’s recap of Markdown’s unlikely ascent) saves notes as individual files with names derived from their titles. But since characters like a colon and a slash are illegal filename bits in many of the operating systems Obsidian supports, this clashed with my habit of including dates in mm/dd/yyyy format in the titles of notes from specific events. Obsidian usually took out the slashes, leaving it to me to insert dashes in their place to restore sense to smushed-together date-digit strings like “6122018,” but in a few cases the Mac app gave up and renamed a note “Untitled.”
    • Mastering Markdown: I feel like I’ve had to reprogram part of my brain to deal with Markdown’s note-title restrictions, but this is nowhere near the first time I’ve done something like that for the sake of note-taking efficiency (for example, the way I hand-write a capital G remains influenced by Palm OS Graffiti). Markdown also uses a syntax of markup characters to format strings of text, which Obsidian hides in practice with the usual keyboard shortcuts and formatting buttons… except when I edit, select or tap inside a word that’s already been formatted, which surfaces the hidden markup characters around it. I find these compromises a reasonable price for the long-term reward of using a lock-in-resistant format that nobody owns and which any text-editing app can read.
    • Writing: Note-taking has always been a text-first experience for me, going back to Palm handheld organizers, and Obsidian has proved pleasantly efficient in that while delivering two unexpected advances over Evernote. One is a tiny bit of input acceleration: When I type a quotation mark or parenthesis, Obsidian automatically adds the closing character in front of the cursor. The other is a feature that’s mandatory in word processors but uncommon in note apps: a word count you can have displayed in the bottom right corner, and which responds to your selection of a block of text by showing the words and characters in that selection.
    • Syncing: The part of Evernote I was most reluctant to leave behind was the synchronization that Bending Spoons had immensely improved, to the point where I accepted a steep rate hike at the end of 2023. Obsidian doesn’t match Evernote’s near-real-time sync, which let me see characters I’d typed on one device appear an instant later in another device’s copy of Evernote. But Obsidian’s optional sync service, $48 a year, is damn fast on its own, generally getting edits from device to device in under five seconds. And because every note is a text file, I can get a note from my phone to my laptop even when both are offline: plug the phone into the laptop via USB, open Obsidian’s folder on the phone in Windows Explorer, and open the file in any text-editing app.
    • Security: Obsidian Sync features end-to-end encryption between devices by default, a security feature I have come to appreciate more after seeing how the Trump administration is willing to treat certain journalists. The tradeoff of that approach and its device-to-device architecture is not having any Web app for Obsidian–but since I almost never used Evernote’s Web app, I can live with that.
    • Plug-ins: Obsidian’s resolutely indie developers tout the ability to extend and augment the app with some of the 5,500-plus plug-ins available. I’m not entirely sold on this tinkering-required approach, since some of these–in particular, the Importer plug-in maintained by Obsidian’s own developers–provide capabilities so essential that they shouldn’t require in-app downloading. In other cases, there are so many plug-ins for a specific task that I’ve postponed picking one. That said, I will put in a plug for the free Notebook Navigator, which lets you switch between notes and folders in a simple tree-style interface like the one I got used to in Evernote.
    • Incomplete items: After succumbing to paralysis via analysis when looking at all of the different audio-transcription options, I opted to outsource that task to a third-party app–the Google Recorder app on my Pixel 9 Pro, which does its AI transcription offline. But I have yet to find any plug-in that can match Evernote’s ability to scan and digitize business cards for easy access to people’s contact details. That is why the desk in front of this keyboard now features a stack of recently-collected business cards two inches tall, held together with an elastic band until I can find some other way to save the useful bits from them in a searchable way. Any suggestions?
    #AITranscription #businessCardScanning #Evernote #GoogleRecorder #handheldOrganizer #interviewTranscription #Markdown #notebook #NotebookNavigator #notes #Obsidian #ObsidianPlugIns #Palm #PalmOS #takingNotes
  3. Google Voice introduced premium personal paid plans starting at $10 monthly. Discover new features like number retention and AI call transcription today.

    #GoogleVoice #Google #Subscription #AITranscription #TechNews

    securityonline.info/google-voi

  4. A lesson in AI transcription: The “play” button in Google Recorder is there for a reason

    A little over two months after I canceled my Evernote subscription instead of eating a 92 percent subscription-rate increase, I’m still figuring out how to handle two tasks that app had handled well. One is scanning business cards, a chore I have set aside because I can still see part of my desktop at home; the other is transcribing recordings of interviews.

    For the shortest conversations, typing in real-time on my phone or laptop suffices. The same goes for interviews with people so practiced at talking to journalists that they habitually speak at a measured pace and in concise sentences.

    But in many other cases, it’s easier to record the conversation and transcribe it later–and over the last several years, machine learning and AI have helped software get good at that. I had used Otter for a while before Evernote added its own implementation and then dramatically improved it; dropping that subscription in favor of Obsidian, which doesn’t seem to offer a comparable tool, required finding a replacement that would ideally come without its own fees.

    Google’s free Live Transcribe is built into Android’s accessibility software and so was my first choice. But I soon got grumpy with its iffy word recognition, clunky controls, and limited export options.

    My next stop was another Google app, this one reserved for Pixel phones like my own Pixel 9 Pro: Google Recorder.

    The single biggest advantage of that free app is that its AI transcription works offline and on-device. That keeps it usable in settings with problematic bandwidth and keeps my recordings private instead of requiring a cloud upload for transcriptions. Recorder also offers multiple export options via its Share menu: saving a .txt transcript file to my phone or other destinations, creating a new Google Docs file, or dumping the transcript into Google’s NotebookLM AI tool.

    But after two months of leaning on this app, I have to report that its accuracy also falls a little short of what Evernote did and what I’ve seen from briefly trying such competing apps as Read AI. Recorder often takes a few words to recognize a change in speakers, it sometimes split sentences when it shouldn’t, it can miss brief interjections, and it incorrectly capitalizes enough nouns to make me wonder if it was trained on German source texts.

    Whatever the faults, those are usually easy problems to fix. I will open that transcript in whatever editing app I prefer, tap the “play” button on the recording on my phone or its cloud backup, then type in whatever Google’s AI missed. This usually requires only one pass, and since I’d want to listen to a recording of another AI transcript before going with its output anyway, it doesn’t feel like time seriously wasted.

    None of this seems seamless. But when you’re trying to turn somebody else’s speech into quotable copy, nothing is–as I was reminded of recently when I opened notes I had hurriedly typed from a meeting that I was sure would be quick enough not to require recording, then wished I had a rewind button on that decision.

    #AI #AITranscription #ces #Evernote #GoogleRecorder #interview #LLM #machineLearning #meeting #MWC #Obsidian #Otter #PixelRecorderApp #ReadAI #transcription
  5. "AI tools are making potentially harmful errors in social work records, from bogus warnings of suicidal ideation to simple “gibberish”, frontline workers have said.

    Keir Starmer last year championed what he called “incredible” time-saving social work transcription technology. But research across 17 English and Scottish councils shared with the Guardian has now found AI-generated hallucinations are slipping in.

    As scores of local authorities begin to use AI note-takers to accelerate recording and summarisation of meetings with adult and child service users, a seven-month study by the Ada Lovelace Institute found “some potentially harmful misrepresentations of people’s experiences are occurring in official care records”.

    The independent thinktank found that one social worker who had used an AI transcription tool to create a summary said the technology had incorrectly “indicated that there was suicidal ideation”, but “at no point did the client actually … talk about suicidal ideation or planning, or anything”."

    theguardian.com/education/2026

    #AI #GenerativeAI #AITranscription #SocialWork #UK #Hallucinations

  6. Plaid just rolled out a new button in NotePin that lets you long‑press a recording, feel a haptic cue, and instantly get AI‑assisted transcription and a concise summary. Powered by generative language models, it turns minutes of audio into readable notes in a tap. Curious how it works? Dive into the details. #Plaid #NotePin #AITranscription #GenerativeAI

    🔗 aidailypost.com/news/plaid-add

  7. Plaud Note Pro is an excellent AI-powered recorder that I carry everywhere

    There has been a flurry of AI voice recording gadgets like Omi, Bee, and Friend that want to…
    #NewsBeep #News #Artificialintelligence #AI #AInotetaker #AItranscription #ArtificialIntelligence #AU #Australia #plaud #recording #Technology
    newsbeep.com/au/381157/

  8. AI transcription accuracy now exceeds 90% on clear English audio, yet professionals still default to keyboards for precision work. The $4.7B voice-to-text market offers conflicting trade-offs: privacy vs cloud accuracy, meeting bots vs local processing, editing time vs typing time. No clear winner emerges in this crowded field. #AITranscription #VoiceTech #WorkplaceTech

    implicator.ai/the-4-7-billion-

  9. #AITranscription : #LimitlessAI has been acquired by #META #Facebook.

    As of December 5, 2025, the service will no longer be available in #Brazil, #China, the #EuropeanUnion, #Israel, #SouthKorea, #Turkey, and the #UnitedKingdom.

    Users in these regions will have until December 19, 2025 to download their information, after which, their account and all associated information will be permanently deleted.

    #transcription #LiveTranscribeApp #transcribeaudio #transcribeapp #Transcribe

  10. Every other transcription app is in the storage business. We're in the forgetting business. Your iPhone remembers so the internet doesn't have to.

    apps.apple.com/gb/app/meeting-

    #PrivacyTech #NoCloud #iPhone #AITranscription #DataSecurity #Meetings #LocalStorage #Tech

  11. Every other transcription app is in the storage business. We're in the forgetting business. Your iPhone remembers so the internet doesn't have to. apple.co/4nDlJvR #PrivacyTech #NoCloud #iPhone #AITranscription #DataSecurity #Meetings #LocalStorage #Tech

  12. Manual transcription is out — AI is in. Discover how transcription tools powered by AI are saving time, boosting accuracy, and transforming audio into text in seconds.
    Full post here:
    #aimartz #aimartz.com #AITranscription #SpeechToText #ProductivityTech

    aimartz.com/blog/transcription

  13. This week, the delightful Adam Engst of TidBITS joins me to discuss the results of his attempt to measure the audio transcription accuracy of several different AI tools. Emphasis on "attempt".

    podfeet.com/blog/2025/03/ccatp

    #AITranscription

  14. @easternblot and others: do you inform your sources about or even ask for permission to use the their audio with AI transcription programs?

    I haven't used those services yet because it feels weird to hand a recording to a third party. I wish there was reliable self-hosted transcription.

    #AI #transcription #AItranscription #otter

  15. Casual app update via #MacAppStore for #AI-Transcription.

    A 5,2 GB update. What the hell am I installing there?

    And how does this go through the App vetting process. I mean, I know quality control is a major issue at
    #Apple but I keep being astonished about their huge and growing list of fuck-ups.

    #itjustworks #macOS

  16. Oh, I think this programmer has similar relationship issues to the others.

    What was said: Sometimes these things get picked up

    What AI said: Sometimes these things hook up.

    #AITranscription

  17. NC #955 Balanced Audio, Whisper Transcription aka MacWhisper, Wahoo Tickr Heart Rate Monitor, Callsheet for TV and Movies podfeet.com/blog/2023/08/nc-95

    Thanks to Jill from the Northwoods, @bbusschots, and Andy from New Hampshire for contributing such great content.

    #cycling #heartrate #AITranscription #moves #TV @caseyliss

Share on Mastodon

Enter the server where you have an account.