An MP3 file is a very poor source for any kind of analysis. This is because MP3 files involve lossy compression, and the degree of compression can vary depending on how the file was encoded. The lossy compression process removes content that the human ear-brain system doesn't notice, but some of that information may be relevant for the type of forensics you are proposing.
However, if you want to do digital forensic analysis on a signal, you will have to work with a digital file of some kind. Ideally, you will work with an uncompressed audio file such as a WAV or AIFF. There are a wide variety of digital signal processes and algorithms out there, so digital is probably the domain you want to work in.
If you want to deal with pure analog, then you have to remain in the analog domain. That probably will require construction purpose-specific equipment.
Is it possible to determine without a doubt that the voice heard there is from the original source
This sounds like a legal definition or other domain-specific definition that you are describing. What does "without a doubt" mean? What does "is from the original source" mean?
Some of this analysis will necessarily be out-of-band. What I mean by that is that a digital file can be duplicated and copied. Certain kinds of filtering can be applied after the fact that could be reasonably expected to have been applied at recording time - such as low-pass or high pass filtering or dynamic range compression. It may be very hard or impossible to prove when this kind of transformation was applied to a signal regardless of whether you are dealing with analog or digital material.