Import the recording from your computer
Save the recording as a supported file such as MP3, WAV, M4A or MP4. Open Audio file in WhisperARC and add it to the queue. The app prepares the media locally before recognition.
Choose Local for on-device transcription or an online source if that suits the recording. Long jobs are divided into sections and joined into one transcript. Keep the source file available while the job is running.
Review the transcript against the audio
Use History and playback to revisit passages you are unsure about. Check names and short negations in the original recording. A clean-looking sentence can still need a correction.
Decide how much editing the reader needs. A verbatim transcript and a summary have different purposes. If you tidy a quotation, keep the meaning and uncertainty that the speaker expressed.
Check speaker labels and timestamps
Speaker labels depend on the selected transcription model and the information it returns. OpenAI's GPT-4o transcribe diarize can supply them. Other models do not automatically identify each speaker.
Use time references to find an important quote again. Some timing information may be estimated when a model does not return detailed timestamps. Review labels and timing against the recording instead of treating them as verified facts.
Export the result as DOCX
Choose DOCX in the export controls and save the document. Open it in Word to adjust paragraphs and add the context the next reader needs. Keep the original recording with your working material so uncertain passages remain traceable.
Imported files are transcribed in their detected source language. Microphone translation, custom spellings and dictation modes do not apply to that file workflow. You can also export TXT for plain text or SRT/VTT when you need subtitles.
