VICIdial audio: GSM vs WAV

The VICIdial audio store takes two formats, and the obvious question is which one to use. The short answer: WAV for anything you want to sound clean, GSM when size matters more than fidelity. Both play fine — the trade-off is space against quality.

For most prompts the difference is small enough that it does not matter. It starts to matter when you have a lot of files, a lot of servers to sync, or a prompt where audio quality carries weight.

What each format is

A VICIdial WAV is PCM mono, 16-bit, 8k — uncompressed audio. Nothing is thrown away, so it is the most faithful version of your recording the line can carry. The cost is size: every sample is stored in full. A GSM file is 8-bit 8k and uses a lossy Codec built for telephony, the same family of compression that low-bandwidth calls rely on. It squeezes the file down hard, at the price of a little quality you usually cannot hear over a phone.

Neither beats the limit of the line itself. A G.711 codec call already tops out at 8 kHz telephone quality, and a G.729 codec leg compresses further, so a pristine WAV and a decent GSM can sound nearly identical to the caller. The gap shows up most on music or long prompts, least on short spoken lines.

How to choose

A simple rule of thumb:

• WAV for short, important prompts — a Welcome message, a legal Safe Harbor message, anything where clarity counts. The size is trivial for a few seconds of audio.
• GSM for long or many files — Music on hold loops, a Soundboard full of clips, archives where the count adds up. Smaller files distribute faster and use less disk on every server.
• WAV as your master, always. Keep the uncompressed original and generate GSM from it if you need the smaller copy. You cannot recover quality from a GSM later.

flowchart TD
  A["New prompt"] --> B{"Quality critical?"}
  B -->|"Yes: greeting, legal"| C["Use WAV"]
  B -->|"No"| D{"Long or many files?"}
  D -->|"Yes: hold music, archive"| E["Use GSM"]
  D -->|"No"| C
  C --> F["Keep WAV master"]
  E --> F

In practice

Most teams settle on WAV by default and reach for GSM only when disk or sync time becomes a real concern. With a handful of prompts on one or two servers, the WAV size is a rounding error and the extra quality is free. When you are running many dialers and pushing a large library out to all of them, the smaller GSM files cut distribution time — that two-minute copy step finishes sooner with less to move. An Agent hears no difference on a typical spoken prompt either way.

Whichever you pick, the format spec is non-negotiable — the store rejects anything outside the two. The accepted file formats post has the exact specs, and the audio prompts and voicemail guide ties the format choice back to how prompts get used across the system.

Format choice is a small optimization on top of a store that has to be wired and in sync to begin with. A managed VICIdial hands you that store ready, so you spend your time picking WAV or GSM, not setting up central sound control. If that sounds easier than building it, spin up a ready VICIdial server — see pricing. A box is live in under 40 seconds.