Section 2.1 has given an overview of the different factors to take into account when choosing a speech recognition tool, namely the speaker dependence model it uses, its possible modes of operation and its ability to adapt to various vocabularies.
The Speech Recognition Control Language was defined, and four available SR solutions were discussed, and their various qualities are compared in Table 2.1.
| Cost | Support | Platform | |
|---|---|---|---|
| Nuance8 | High | very good for customers | Windows, Linix |
| IBM ViaVoice | Medium | developer docs, mailing lists | Linux, Windows* |
| XVoice | free (requires ViaVoice) | minimal | Linux, Unix |
| CVoiceControl | free | minimal | Linux, Unix |
As Table 2.1 shows, of the four technologies discussed, Nuance would be the first choice should money be no object. Of the free applications, IBM ViaVoice is at the top of the pile due to the large amount of documentation, and searchable mailing lists that are available online. Since XVoice requires ViaVoice in order to function, there seems little point in using it unless you have an existing application, or applications, written for the X windowing system for which you require speech access to. CVoiceControl offers it's own simple recognition engine making it a complete speech recognition package. However, it only uses isolated word recognition, making the scope it can be applied in rather limited.