Audio transcription — how to use it in the Studio
Send an audio file and get the transcription ready: plain text, dialogue split by speaker (diarised) and the detected audio events. The transcription comes in the same language as the audio.
The best path: record the call
The recording comes from the device itself (when the model allows it) or from the phone service (VoIP/PBX).
On the phone (when the model allows it)
Redmi / Xiaomi (MIUI or HyperOS)
Open the Phone app (Xiaomi's, not Google's) → ⋮ → Settings → Call recording → choose Record always or contacts only. The audio is saved as .m4a (MIUI/sound_recorder/call_rec folder).
Samsung (One UI)
Open the Phone app → ⋮ → Settings → Call recording (available on some models/regions). If it is not there, look for the Record button on the ongoing call screen. The audio is saved as .m4a.
Other Android
Open the Phone app → Settings and look for Call recording. If your model does not have it, use the service recording (next to it).
Through the phone service (companies)
IP PBX (FreePBX/Asterisk, 3CX, Grandstream)
Enable recording on the extension or trunk. The file stays on the server (.wav or .mp3) and you download it by period. Best option for companies.
Teams, Zoom and Google Meet
Use the Record button in the call or meeting (the app warns the participants). Then download the file (.mp4 or .m4a).
WhatsApp (calls)
WhatsApp does not record calls by itself — use the device recording (above) or the service one.
Step by step in the Studio
- In the FalaAI Studio, Input tab, Audio card: drop the file on the dashed area or click it to choose (you can pick it straight from Google Drive).
- If you want, open the CFG and fill in the Reference (optional) — then click TRANSCRIBE.
- The time depends on the audio size. When it finishes you get the text, the dialogue formatted by speaker and the detected audio events.
- Save the HTML report (Open HTML / Download HTML) — that is your evidence.
- Then generate the Diagnosis and the Audit. The details of each one are in their own Diagnosis and Audit help pages.
What you need to know
What it costs
- 1 credit per second of audio.
- Minimum of 10 credits per file (an audio of 10 seconds or less costs 10).
- Examples: 1 minute = 60 credits · 10 minutes = 600 credits · 1 hour = 3,600 credits.
- The total is added up exactly, with 2 decimal places.
If something goes wrong
- Empty file: choose the audio again (it may not have uploaded completely).
- Format not recognised: use .mp3, .mp4, .m4a, .wav, .ogg or .opus.
- File too large (over 1 GB): split the audio or export by period.
- Audio too long (over 3 hours): split it into parts of up to 3 hours.
- The audio did not open: it may be corrupted — record or export it again.
- The transcription failed: try again; if it keeps happening, contact support.
The decision to record is yours and the legality depends on your country and the consent of the people on the call. Audio is personal data (GDPR/LGPD) — here we only teach you how to do it.