Multi-Language Transcription

Supported Languages

NoteLLM supports transcription in four languages:

• English -- full support in both local and cloud recognition modes.
• Chinese (Mandarin) -- full support in both local and cloud recognition modes.
• Japanese -- full support in both local and cloud recognition modes.
• Korean -- full support in both local and cloud recognition modes.

You can switch between languages before or between recordings. The app applies the selected language to the speech recognition engine so it knows which phonetic rules and vocabulary to use.

How to Select a Language

1. Open NoteLLM on your device.
2. Before starting a recording, tap the language selector near the record button.
3. Choose your desired language from the list.
4. Start recording. The transcription engine will process your speech in the selected language.

The language setting persists between sessions, so you only need to change it when switching to a different language.

Transcription Accuracy

Transcription quality depends on several factors:

• Recognition mode. Cloud recognition generally delivers higher accuracy than local recognition, especially for complex vocabulary, technical terms, and accented speech. See Recognition Modes for a detailed comparison.
• Audio clarity. Clear speech in a quiet environment produces the best results. Background noise, crosstalk, and muffled audio can reduce accuracy.
• Speaking pace. A natural, steady pace works best. Speaking too quickly or mumbling may cause words to be missed or misrecognized.
• Language consistency. For the best results, speak primarily in the selected language. Code-switching (mixing languages mid-sentence) may reduce accuracy.

Editing Transcriptions

After a recording is transcribed, you can edit the text directly within NoteLLM. This is useful for correcting any recognition errors, adding punctuation, or restructuring sentences. Edits are saved automatically as part of the note.

See Note Management for more details on editing and organizing your transcribed notes.

Tips for Better Transcriptions

• Choose the correct language before recording. Mismatched language settings will produce poor results.
• Use cloud recognition when accuracy matters most, such as for important meetings or detailed notes.
• Use local recognition when you need offline access or want maximum privacy.
• Speak clearly and at a natural pace. Short pauses between sentences help the engine segment text accurately.