Have something to say?
Trigger transcription of existing recordings by ID
Hi, I'm using MeetMic with the local API and an Obsidian integration. I often have back-to-back meetings, so I've disabled automatic transcription to avoid interfering with recording. I'd like to automate transcription of existing untranscribed meetings at the end of each day, processing them sequentially. I can identify untranscribed meetings using GET /api/v1/transcriptions/:id and isTranscribed: false, but I couldn't find an API or CLI command to start transcription for an existing recording by ID. Is there a supported way to trigger the native Transcribe action programmatically, without reimporting the audio or creating duplicate entries? Thanks!
Allow returning to main UI during transcription/speaker detection
Right now, when transcription or speaker detection is running, you can’t go back to the main UI. If you leave the view, transcription/speaker detection is canceled. It’d be great to be able to navigate back to the list, view other transcripts, etc. since transcription/speaker detection can take very long.
Word-Level Timestamp Option
It would be awesome to have word-level accuracy for the timestamps. For some in research, this is very important, to improve annotation accuracy.
Return the "New Audio" Button with System Audio Toggle
The fact that there is no option to record system audio means it won’t record system audio. This is a major problem and it completely loses its utility without it. Also, it frequently crashes now on 1.27
Project folders or tags
I have too mane transcriptions. Finding right one is a challenge sometimes. Batch transcriptions may include 100+ files. This floods sidebar quickly. Adding some grouping tools to the desktop app would help: Project folders to collect recordings relevant to particular project Tags to filter by company name, type of recording (like meeting, interview, important), speaker etc.
Floating meeting recorder view options
When you start a meeting recording (via meeting detection), the floating recorder stays in place. An option to hide that floating thing completely or move that to the menu bar would be really nice.
Speaker name autocomplete
Right now, you have to type out the full speaker name. It’d be great if speaker names can be autocompleted from previously used names.
`name` field empty in full transcription API endpoint
The /api/v1/transcriptions/:id detail endpoint returns an empty string for name, while the list endpoint (/api/v1/transcriptions) returns it correctly. List endpoint: "name": "New Project (9:06 PM)" Detail endpoint: "name": ""
Segments missing speaker field in transcription detail endpoint
The /api/v1/transcriptions/:id endpoint documentation shows a speaker field on each segment, but the actual response doesn't include it. Documented: { "id": "seg_001", "start": 0.0, "end": 4.52, "text": "...", "speaker": "Speaker 1" } Actual: { "id": "AFF66A02-...", "start": 13280, "end": 232080, "text": "..." } Also noticed segmentCount, isTranscribed, and per-tag color from the docs are absent in the actual response. The segment start/end values appear to be integers (milliseconds?) rather than the floats (seconds?) shown in the docs.
Most recent transcript API endpoint
Love the local server API endpoints. I’ve been eyeing this app and purchased it today after seeing the 1.25 changelog. Trying to create some automation using the local API server, and thought it’d be great to be able to fetch the last transcription without having to list all transcriptions and find the ID of the most one.
Diarization Model Selection
I would love to see other ways of automatically diarizing the transcripts. The current implementation works if the audio is great, however there are other approaches existing already (e.g., https://huggingface.co/nvidia/diar_sortformer_4spk-v1). It would be nice to be able to switch between models, similar to the way one can switch the models for transcribing locally.
Improve Reliability of YouTube Video Downloads
Currently, our YouTube download functionality is inconsistent due to frequent, unannounced changes to Google's internal APIs and player structures. This causes download failures that frustrate users. This initiative focuses on re-architecting our downloader to handle these upstream changes more gracefully.
Translations
I would suggest feature, after transcribing, to get with one click translation to other language. even to have 2 column layout side by side to see original and translation texts. I like that Turboscribe has similar thing, so we can learn foreign languages that way, by transcribing online videos. enable link transcriptions (Facebook, Instagram, Youtube, X, Rumble) if not yet. maybe even allow export to some google sheets and some flashcard software (desktop ones), to have table with original words and verbs, and translations. So that we can print it, and learn language (without repeating words on ourself, nothing stays in head).
iOS version
As a user of iOS, I want to record in-person meetings and voice memos on my mobile device, so that I can transcribe and save them without using my Mac.
Enhance dictation quality with AI
Use LLM models to improve punctuation and grammar of dictation segments
Watch folders
Add the ability to designate folders for automatic monitoring. When new audio or video files are added to a watched folder, the app will automatically transcribe and translate them in the background.
Calendar integration for meeting recordings
As a user, I want the app to integrate with my calendar so it can schedule/auto-start recordings for my meetings and use the meeting details (title, participants, links).
Improved support for USB microphones and diverse audio devices
External microphones (USB, audio interfaces) expose varied sample rates, channel counts, bit depths, and clock behaviors. Current capture is tuned to built-in Mac mic defaults, causing incompatibilities, resampling artifacts, device init failures, or silent tracks when device capabilities don’t match assumptions. We need to reliably capture audio from a wide range of USB and external microphones.
Preferred audio track selection for YouTube downloads
Enable users to select a preferred language of the audio track when downloading from YouTube so the captured audio matches their intended language.
Audio meters in Recording panel
During meetings, users can’t easily tell whether their microphone and system audio are being captured. Missing OS permissions or incorrect input/output settings can result in silent recordings without clear, real-time feedback. Display real-time audio level meters for microphone and system audio in the recording panel so users can quickly verify signal presence, detect silence/clipping.