# Recording & Transcription

Use the Voice API v2 to record and transcribe your phone calls. You can choose to record the entire call or isolate specific call legs to capture exactly what you need. By default, the Voice API records all legs in a call.

Recordings give you dynamic control over your audio capturing. You can start and stop recording at any given time during an active call. Once the call or recorded segment is complete, the API automatically generates a post-call audio file.

## Features

Flexible recording scopes: Record all legs in a call (default) or target individual call legs.

On-demand control: Start and stop recording at any given time during a call.

Post-call file generation: Automatically generate an audio file after the call completes.

Format options: Save your recording files in WAV or MP3 formats.

Post-call transcription: Automatically convert your recorded audio into text. Call recording must be enabled to use transcription features.

## Post-call transcription

You can automatically transcribe your finalized recordings into text for analytics, compliance, or record-keeping.

The transcription engine supports over 80 languages and regional locales globally. To ensure high accuracy for local accents, it includes extensive support for regional varieties of major global languages—such as distinct geographic locales for English, Spanish, French, Portuguese, Arabic, and Chinese—alongside dozens of other European, Asian, Middle Eastern, and African languages.

## Developer tips

Protect sensitive data: Use the dynamic start and stop controls to temporarily pause recording when users provide sensitive information, such as credit card numbers or PINs, to maintain compliance.

Optimize your file format: Choose MP3 to minimize file size and storage costs for long-term archiving. Choose WAV if you require uncompressed, maximum-quality audio to improve the accuracy of post-call transcription.

Isolate legs for analytics: If you plan to run sentiment analysis or speaker-specific transcriptions, record by individual call leg rather than recording the entire call. This gives you isolated audio tracks for cleaner processing.

## Common pitfalls

Instant availability assumption: Post-call file generation happens asynchronously after the call ends. Do not attempt to fetch the file immediately upon call disconnect; rely on webhook callbacks to notify your application exactly when the recording is ready.

Missing transcription prerequisites: Because transcriptions are generated from the audio file, attempting to run post-call transcription without explicitly enabling recording on the call or leg will result in an error.

Overlooking the default scope: Remember that the default setting captures all legs. If your use case requires recording only the customer (and not the agent or IVR prompts), you must explicitly set the scope to that specific call leg.