Tandem signaling
From Wikipedia, the free encyclopedia
Tandem signaling is one of the most difficult conditions for speech coders to perform well in is the case where a digital speech-coded signal is transmitted from the mobile to the base station, and then demodulated into an analog signal which is then speech coded for retransmission as a digital signal over a land line or wireless link. This situation, called tandem signaling, tends to exaggerate the bit errors originally received at the base station. Tandem signaling is difficult to protect against, but is an important evaluation criterion in the evaluation of speech coders. As wireless systems proliferate, there will be a greater demand for mobile-to-mobile communications, and such links will, by definition, involve at least two independent, noisy tandems.
[edit] Tandeming in the Real-World
For example, if two individuals are using two different mobile phones on two different networks, and attempt to call each other, Tandeming will occur. This is even more evident - in terms of speech intelligibility, when one user is using a phone with a different audio codec than the other user. If one user has a high-bitrate audio codec, and the other uses has a low-bitrate, the user with the high-bitrate will be able to hear the user with the low-bitrate, as the tandeming will not have enough of an opportunity to make the speech completely unintelligible.
However, if two users on two different mobile phone networks, with different audio codecs [Such as AMR-Narrowband, and EVRC] have their codecs falling back to the lowest possible birate on each end [4.75, ~6 kbit/s], then the speech will suffer from problems. Tandeming is nearly impossible to avoid, because even if all mobile phones used one audio codec, there is the issue of packet retransmission versus re-encoding. When a user makes a call to a land-line phone, the packets "flow" to the tower in that particular cell, and are then decoded into 64 kbit/s PCM. This creates little to no extra noise, and if the signal is sufficient enough, the user on the land-line will be able to carry on a conversation. If, however, the user on the mobile phone calls another mobile phone, the packets cannot simply be 'shifted' to the other user's mobile phone. Instead, they must be sent to the tower, decoded into PCM, sent to the other corresponding tower, and then transcoded to the audio codec used on the other handset. This results in an obvious loss of speech quality the moment the other uses picks up the phone.
Tandeming can be demonstrated using a Personal computer, a microphone, and Apple's Quicktime Pro software, version 6.5 or above:
- Record a fictional conversation with your microphone, ensuring that the audio levels are sufficient, are not too high.
- Save the file was a WAVE, AIFF, or any uncompressed format.
- Open the file in the Quicktime player.
- Choose export to Quicktime movie, and choose AMR-Narrowband at 12.2 kbit/s, 8 kHz, mono.
- Give the file a name, export the file.
- Open up the exported file, and export another file. This time, change the audio codec to Qualcomm Pure voice, 8 kHz, mono, half-rate, optimize for streaming.
- Give the file a new name, then save it.
- Compare the two files.
Using this simple tandem experiment, you can actually hear what goes on in the real-world for mobile phone users day to day. To the human ear, there will be a discnerable loss of audio quality. [It is to be noted, however, that the degree to which any given human ear will be able to determine said loss of audio quality is subjective and varies substantially.]
