TotalApp Docs
Ctrl K
EN | TR
Go to App

Speech to Text

Turn spoken audio into text. Record from your microphone or upload an audio file, pick the language (or let the model detect it) and copy the transcript into your work.

Overview

Speech to Text is a tool inside Audio Creator. It transcribes what is said in a recording using a Whisper model hosted on the Modelion gateway, and shows the result in a box you can copy. Nothing is installed: the audio is sent through the TotalApp server, which holds the API key, so the key never reaches your browser.

Key Capabilities

  • Record directly from the microphone, with a live waveform while you speak
  • Or upload .mp3, .wav, .m4a or .webm files up to 25 MB
  • Preview the clip before transcribing
  • Choose the spoken language or keep Auto-detect
  • Copy the transcript with one click

How to Transcribe

1. Open Speech to Text
→
2. Record or Upload
→
3. Pick the Language
→
4. Transcribe
  1. Open Audio Creator and click the Speech to Text card.
  2. Click Record and speak; the bars move with your voice. Click Stop when you are done. Or click Upload audio file and choose a file.
  3. Play the clip back to check it. Use the × button to discard it and start over.
  4. Choose a Language from the list, or leave it on Auto-detect.
  5. Click Transcribe. A spinner shows while the audio is processed. The transcript appears below; click Copy to put it on the clipboard.

Choosing the Language

With Auto-detect the model guesses the language from the audio. This works well for clear, longer speech, but short, quiet or noisy clips can be mis-detected (for example, Turkish speech returned as Korean text). When that happens, pick the language explicitly. The list covers Turkish, English, German, French, Spanish, Italian, Portuguese, Dutch, Russian, Arabic, Persian, Hindi, Chinese, Japanese and Korean.

Recording quality

Recordings made in the browser are converted to 16 kHz mono WAV before upload, which is the format the speech model expects. For the best accuracy, record in a quiet room, keep the microphone close, and speak in one language per clip.

Limits and Troubleshooting

SituationWhat to do
"Microphone access was denied or is unavailable"Allow microphone access for the site in your browser's address-bar permissions and try again.
"The file is larger than 25 MB"Trim or compress the audio, or split it into shorter clips.
"Unsupported file type"Use an .mp3, .wav, .m4a or .webm file.
The provider account is out of creditAsk your administrator to top up the AI provider account, or try again later.
Wrong language in the transcriptSelect the language manually instead of Auto-detect.