3.0

Algorithms for timestretching/pitchshifting

Standard: Timestretching and pitchshifting in standard quality. This method is suitable for audio
material without a pronounced beat. Beat markers are evaluated to improve audio quality.

Smoothed: Timestretching and pitchshifting for audio material without pulsing elements. The
method is suitable for polyphonic orchestral instruments, pauses, speech, and singing. Beat
markers are not evaluated.

In this case, a considerably more complex algorithm is used which requires more processing time.
The material can also be processed with very large factors (0.2... 50) without causing serious
artifacts. The material is "smoothed" to make the sound softer and emit it at an adjusted phase
level. This smoothing is hardly audible, for example, with speech, singing, or solo instruments.
Problems may arise with more complex spectra (sound mixes from various instruments or
finished mixes). For smaller corrections (factors approx. 0.9... 1.1), we recommend setting the
smoothing level as low as possible.

Beat marker slicing: Beat-synchronous timestretching and pitchshifting via splitting and
temporal repositioning. Exactly set beat markers are required at the beats or transients. The
markers can be generated in real time (automatically) or read from the WAV file if available
(patched). In the deluxe version's included MAGIX Music Editor, a patching tool is provided for
users to set the markers themselves. The algorithm is suitable for rhythmic material that can be
divided into individual beats or notes. This requires a low audio level before each beat or note.

Beat marker stretching: Beat-synchronous timestretching and pitchshifting in standard quality.
The material is stretched between beat markers positions so that the beats or attacks at the beat
marker positions are not impaired by stretching. The markers can be generated in real time
(automatically) or read from the source file if available (patched). This method is suitable for
rhythmic material that can not be divided into individual beats or notes because the beats or notes
overlap each other.

Universal HQ: Universal methods for timestretching and pitchshifting in very high audio quality.
Suitable for all types of audio material. Beat markers are evaluated to improve audio quality. This
method requires a lot of time for processing, so application of the "Calculate all real-time effects"
function ("Edit" menu) is recommended.

Monophonic voice: Timestretching and pitchshifting for vocal solos, speech, or solo instruments.
The material must not contain background noise, and excessive reverb may also be detrimental to
the use of this method. With suitable material the audio quality is very high. The "Correct formant
factor" option preserves formants if pitches are changed. These are characteristic basic
frequencies of the voice that are independent of the pitch that is sung. In other words, the
characteristic discoloration of pitch ("Mickey Mouse") effect does not occur in this case. The
formants, however, can be shifted by +/- 12 half tones. This achieves suitable vocal distortions.
Beat markers are not evaluated.

Resampling: Pitch shift and tempo cannot be changed individually. This method requires
considerably less CPU time. If the pitch is increased or the sample is shortened, then resampling
is almost completely free of loss, and the sample material will suffer almost no damage. In other
cases, resampling causes loss of overtones. For example, if the length of a 44.1 kHz sample is
doubled, then the freqency level of the result will be limited to 11.025 kHz. The sound is the
same as when the playback speed of a record player or tape recorder is changed.
Page 107