Support

Record an idea

1. Tap the microphone and speak. With “Record immediately” in Settings, recording starts as soon as you open the app.
2. The live text shows what your iPhone understands.
3. Tap stop when you are done. Murmel turns it into a note.
4. To speak the next idea right away, tap “Hand off” instead of stop. The recording is finished in the list.

Check and save the note

1. After you stop, Murmel shows the note with title, category, due date and tags.
2. If everything is right, tap “Save”.
3. Otherwise tap “Edit” and change the fields.
4. If Murmel misheard a word, correct it in the note’s transcript. Title and note can then be created again from it.

Organise with tags

1. In a note, type a new tag under Tags and choose “Add”.
2. Settings > Tags lists all tags. Murmel suggests adding a tag to earlier notes that mention the word.
3. Use the filter in the list to filter by tag and category.

Record on Apple Watch

1. Open Murmel on your watch, from the complication or the button in Control Center.
2. Speak and stop. The recording goes to your iPhone and becomes a note there.
3. If your iPhone is out of range, the recording waits on the watch. “Transfer” shows what is waiting, and “Send now” pushes it.

Sync with Apple Reminders (Pro)

1. In Settings under Notes, turn on “Save to Reminders” and allow access.
2. Murmel creates one list per category: “Murmel Tasks”, “Murmel Reminders” and “Murmel Ideas”.
3. A checkmark syncs both ways, in Murmel and in Reminders.

Set up your own server (Pro)

With Murmel Pro, your own Whisper server transcribes instead of the iPhone. That pays off with dialects such as Swiss German. Murmel works with any server that offers the OpenAI transcription API, for example speaches with the faster-whisper-large-v3-turbo model. The recording goes only to that server, together with the language you chose in Murmel.

1. On a computer with Docker, create a compose.yaml file like the one below. Next to it, create a .env file with one line API_KEY= followed by a long random value, for example from “openssl rand -hex 32”.
2. Start it with “docker compose up -d”. Load the model once: curl -X POST -H "Authorization: Bearer YOUR_API_KEY" http://localhost:8000/v1/models/deepdml/faster-whisper-large-v3-turbo-ct2
3. In Murmel, open Settings > Your own server and tap “Add server”. Enter a name, the address and the value of API_KEY as the token, then choose the model and save.
4. Under Settings > Recording > Transcribe, choose the server.
5. On your home network, enter the computer's LAN address, for example http://192.168.1.20:8000. To reach the server on the go, it needs an https address, for example through a Cloudflare Tunnel or a reverse proxy. Murmel then shows something like https://whisper.example.com.
6. The computer should have 4 cores with AVX2 and 8 GB of memory. If Murmel rejects the token, API_KEY and the token in the app do not match. If Murmel shows no model, the model has not been loaded on the server yet.

services:
  speaches:
    image: ghcr.io/speaches-ai/speaches:latest-cpu
    restart: unless-stopped
    environment:
      API_KEY: ${API_KEY}
      WHISPER__TTL: "-1"
    ports:
      - "8000:8000"
    volumes:
      - hf-hub-cache:/home/ubuntu/.cache/huggingface/hub
volumes:
  hf-hub-cache: