Your question is a little difficult to answer, simply due to the number of variables. But here is a simplistic answer.
One of the “drawbacks” that is often pointed out, is that the IP network has not resend capability. In most data network protocols, there are provisions for resending lost packets of data. When IP was “invented”, it was decided that for simplicity, that would be eliminated. Thus, packets of anything on the web thsat go astray are lost for ever.. Another stated “drawback” is that IP does not used a fixed point to point connection. Packets can be sent in varying paths, which leads to latency (delay) being introduced. Either of these would degarde the voice being sent.
From a practical viewpoint, in most developed countries the IP networks are being expanded to carry the traffic demands. Having enough network (“pipe”) mitigates or eliminates lost packets and latency. Users are successfully using IP for voice today with very good results. There may occur an occassional network problem, bit those which are obvious to the users are not common.
There are problems with related systems and especially voice mail. In voice being sent over data networks (IP, Frame Relay or ATM) the voice is compressed from its native 64Kbps bandwidth to something more manageablem typically 16 KBps or less. Similarly woice mail systems compress voice for space reasons. At times this dual compression (the network and then a voice mail system) will degrade the voice to an unintelligible garble.
The same can be true for multi-nodal networks. If the voice has to be restored to its analog format and then digitized to be sent again, it can introduce degrading of the signal. Fortunately, most newer router are intelligent enough today to send packets destined for other addresses without doing the digital-analog-digital conversion. However, the network folks have to be smart enough (don’t make that assumption) to ensure that the routers are programmed to automatically forward those packets.
Bottom line, IP of voice can be a very real and viable alternative for businesses. While the voice will not be of the same quality as a digital voice network (due to the compression), many find that its price performance make it a satisfactory alternative. As IP is generally available everywhere (the same cannot be stated for Frame Relay or ATM), it is both accessible and cost effective.
In the future IP will likely become usage priced, as the traditional telephone companies can rightfully state that their cost/pricing structure is based on random calling patterns and not nailed (full time) connections as used in the IP world. There are other regulatory issues also involved that have been raised to the FCC, but not yet acted upon. We all shall wait to hear what the bureaucrats decide. But the prognosis is the IP will become more expensive if and when it has to account for its usage in the local telephone network.
Good luck