Linear-prediction diphone-concatenative synthesis

This use of the LP model leads us to the technique called concatenative synthesis. Here, no explicit vocal-tract modelling is performed rather, an exhaustive set of units is acquired and synthesis is performed by concatenating these units in appropriate order. The targets and transitions are stored within the units, and, so long as suitable units and good original speech are used, it should be possible to generate reahstic parameter patterns via the concatenation technique. 

Synthesis techniques based on vocai-tract models 

For convenience, we often denote a single diphone by both its half-phones joined by a hyphen. Hence the phone string /h eh 1 aw/ gives the diphone sequence /h-eh eh-11-ow/ or /h-eh/, /eh-1/, /1-ow/. 

A key task is carefiil determination of the fiiU set of possible diphones. This is dealt with in full in Section 14.1, but for now it suffices to say that we should find a database in which there is a high degree of consistency, to ensure good continuity and joins, and a broad range of phonetic contexts, to ensure that all diphone combinations can be found. While it is possible to use existing databases, it is more common to design and record specifically a speech database that meets these particular requirements. 

Once the diphone tokens have been identified in the speech database, we can extract the LP parameters. This is normally done with fixed-frame analysis. First we window the section of speech into a number of frames, with, say, a fixed frame shift of 10 ms and a window length of 25 ms. The coefficients can be stored as is, or converted to one of the alternative LP representations. When storing the units in the database, it is important that the durations and FO values of the two half-phones are recorded, since these are needed in the process of prosodic modification. 

---