PtFisch
commented
6 years ago I solved issue 1: There were still remains from my old Jasper-installation (Main-Branch). I uninstalled Sphinxbase and Pocketsphinx in pip and fully removed my old Jasper-folder. Now everything works fine, when starting directly from the Pi console.
I am still eager for input on issue 2 and 3.
G10DRAS
Hello and thanks for bothering with this issue(s):
1) I installed Jasper somewhat successfully (using the dev-branch) with Pocketsphinx (5prealpha) for STT and Flite for TTS. Now I startet Jasper via SSH and everything worked as expected, however, when starting Jasper directly in the Pi-Console (raspbian stretch with Desktop) the debug information tells me that Pocketsphinx is not installed:
Is this in any way normal? If yes, how is it explainable and is it possible correct that?
2) While installing Jasper and doing some research, I often read, that audio samples for STT were recordet at 16kHz, making 44.1kHz and 48kHz audio input rather inaccurate. Is it possible to remedy that without buying new hardware? If yes, how (e.g. code) and where (e.g. precise code part of a file). Are there other steps to take, to optimze Pocketsphinx, like using longer keywords. (Running Jasper completely offline is important to my projects)
3) The Tutorial explaines, how to create Wordfile, pronunciation dictionary and language model. I am uncertain how to achieve the same with Jasper. There are (again, dev-branch) the jasper/data/standard_phrases (not containing "JASPER"). So which files do I have to edit in which way, to change the keywords and further the welcoming-message ("How can I be of service") if this is also possible. What do I need to pay attention to, when creating a new model?
I would really appreciate, if someone is able to answer a question or having some advice. Thank you in advance!