Recognition Methods

Grammar-based

Grammar-based integration involves using predefined grammars or language models to constrain the recognized speech to specific phrases or structures. This method is suitable for applications where the expected input is limited to a predefined set of words, phrases, or commands. The grammar specifies the vocabulary and the possible combinations of words or phrases that the speech recognizer should recognize. By constraining the recognition to a specific grammar, the system can achieve higher accuracy and faster processing.

Dictation

In dictation mode you need to use a language model provided Sestek that contains everything you need in order to perform a successful recognition. It allows for free-form speech recognition, where the system transcribes spoken words into text without specific constraints or predefined grammar. This method is suitable for applications that require transcribing spoken language without restricting the vocabulary or the sentence structure. Dictation integration is often used in applications like transcription services, voice assistants, or voice-enabled text input.

How to choose the suitable recognition method for your needs?###

Despite its advantages over grammar recognition, the dictation mode has a higher error rate, and it is observably slower. Because of different characteristics, the two are not used interchangeably. Both have their own usage areas. Dictation is more suited to systems with large word sets where the order of words is mostly unpredictable. An example might be medical report dictation, IVR natural language processing and agent systems etc.