Hi, i am new with cmu sphinx. I did exactly as the website says to install and configure pocketsphinx with sphinxbase, but when i use it is really weird. If i give it a file, it is not accurate at all, and speaking from the mic has the same result. Also, the "output" comes in a kind of weird way as it starts listening and stops all the time, like a loop. I provide an example, when i try to give it a wav file which says something simple like "read my lips" the output is something totally different.
This wav says: "My biggest job is to prevent the enemy from hitting us again".
This audio has very bad sound quality, it contains reverberation, clipping noise, reduced bandwidth to 5khz. It is not easy to recognize such samples, you have to build a specialized system for this. Or you need to find a way to recieve more high-quality audio.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Ok, i managed to find something better, but it understood only 2 words of 10. I have two questions. Is this the normal output?(kind of messy), and how can i improve the quality of the sounds i give to it?
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
It is hard to give you advise on accuracy without seeing the sample in question. If you want high quality audio example, take an audiobook from librivox.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hi, i am new with cmu sphinx. I did exactly as the website says to install and configure pocketsphinx with sphinxbase, but when i use it is really weird. If i give it a file, it is not accurate at all, and speaking from the mic has the same result. Also, the "output" comes in a kind of weird way as it starts listening and stops all the time, like a loop. I provide an example, when i try to give it a wav file which says something simple like "read my lips" the output is something totally different.
This wav says: "My biggest job is to prevent the enemy from hitting us again".
pocketsphinx_continuous -infile converted.wavAny help appreciated!
You could provide the file
It is just an example. i randomly found it in wavsource.
Last edit: Telis Papageo 2016-05-17
This audio has very bad sound quality, it contains reverberation, clipping noise, reduced bandwidth to 5khz. It is not easy to recognize such samples, you have to build a specialized system for this. Or you need to find a way to recieve more high-quality audio.
Ok, i managed to find something better, but it understood only 2 words of 10. I have two questions. Is this the normal output?(kind of messy), and how can i improve the quality of the sounds i give to it?
It is hard to give you advise on accuracy without seeing the sample in question. If you want high quality audio example, take an audiobook from librivox.