Implementation guides for Voci tuning capabilities like substitutions, redaction, custom language modeling, and hinting

You can use one of Voci's standard language models without any tuning. This option offers world-class accuracy that serves many of Voci's customers well. This option works best with high quality audio. In the context of automatic speech recognition, high quality audio has minimal background noise, is recorded at a bit rate of 64 Kbps or higher, and uses two-channel audio capture for audio with two speakers.

From this strong baseline it is possible to improve accuracy further by tuning Voci's ASR offerings to better understand each enterprise's specific telephone call interactions with their end customers. Voci offers a variety of methods to improve accuracy, some of which can be performed in the field with minimal training and assistance required from Voci's support staff.

These options provide different kinds of accuracy improvements at different levels of effort and cost.