NY6 User Manual
Ver 1.3 2019/03/28
27
2.14.1 Speech Synthesis
NY6 supports 10-bit PCM and encoded 4-bit / 5-bit mixed ADPCM speech data. The PCM voice has
higher quality, but it occupies double ROM space at least than ADPCM. By cooperating with embedded
noise filter of 250KHz over-sampling, it could decode high fidelity voice data even if you adapt ADPCM
voice. It means you could store longer voice duration or provide more kinds of patch at lower sampling
rate but enrich user’s applications without degradation of sound quality.
2.14.2 MIDI Synthesis
There are three combinations to form a patch in NY6. The first way (called Head-Only) is to record a
complete waveform, then play it by playing whole wave only. This is the best way to represent a high
quality patch, but the price has to pay is the ROM cost. In contrast, users can extract the periodic part of
a patch (called Tail-Only), then play it by playing the periodic wave repeatedly. The ROM occupied by
this kind of patch is minimal; however, sound quality is sacrificed.
The compromise architecture is “Head+Tail” with envelope information, which is called ADSR. During
MIDI synthesis, the Head wave is played only once and the Tail wave is always repeated to generate
the synthesis output. Generally, the Head wave is used to represent the non-regular part at the
beginning of a patch or to represent a whole of general voice or sound effect. The Tail wave is to
represent a periodic cycle in the regular and periodic part in a patch. The Head wave and Tail wave are
usually extracted from the same waveform and Tail wave is immediately successive to Head wave. This
patch synthesis method can dramatically reduce ROM size needed to store the patch data.
Besides, a hardware circuit of automatic Tone-Calibration is built-in. It can result in near-zero frequency
deviation for precise generation of tone frequency.
Note: There is a limitation about Tail waves that sample number of Tail waves must be integer
multiple of 4 or 5 according to 5/4-bit data compression of ADPCM.