Voice synthesis on ISR

Page 21/28
14 | 15 | 16 | 17 | 18 | 19 | 20 | | 22 | 23 | 24 | 25 | 26

By [WYZ]

Champion (417)

[WYZ]'s picture

10-02-2017, 22:52

FYI:

Test 1

Test 2

By ARTRAG

Enlighted (6276)

ARTRAG's picture

11-02-2017, 10:06

they sound well

By Metalion

Paladin (1013)

Metalion's picture

11-02-2017, 10:42

Amazing ...
Do I understand well, those SFX are generated only by the PSG using a WAV sample ?

By [WYZ]

Champion (417)

[WYZ]'s picture

11-02-2017, 14:14

Metalion wrote:

Amazing ...
Do I understand well, those SFX are generated only by the PSG using a WAV sample ?

Right, the tool works as a "mp3" for 8bits soundchips. Most of the wavs do not need pre-treatment but for more complex samples the wav must be simplified. The result is quite good, also with only 2 PSG channels reach a very good quality.

By syn

Paragon (1920)

syn's picture

11-02-2017, 14:18

I have been following this topic with half an eye, I thought it was some speech generator but it is actually a sample replayer? how cool Big smile

By NYYRIKKI

Enlighted (5396)

NYYRIKKI's picture

12-02-2017, 10:36

syn wrote:

I have been following this topic with half an eye, I thought it was some speech generator but it is actually a sample replayer? how cool Big smile

No, you can't call this a sample replayer either... it is like calling circle a line... maybe, but you are viewing it 90° wrong in a 3D world. Smile ... CIRCLE (128,106),80,,,,1E-15

Sample players try to reproduce the sampled result, this doesn't. This plays simplified representations of voices that make up the sound in a sample. It is a frequency domain player that takes sample as an input. While sample is a representation of time and volume this is representation of frequency and volume.

By giangiacomo.zaffini

Champion (267)

giangiacomo.zaffini's picture

12-02-2017, 11:29

So, does it select two or three (or more if SCC us envolved) more powerful frequencies and sample their power in time? Cool but I don't think it has an IFFT to return in time domain.

By hit9918

Prophet (2868)

hit9918's picture

12-02-2017, 16:18

is it samples or not.
from point of view of 44khz hifi it is no samples, while from an 8bit point of view it is "wow, sampling the NES".
there was a nice example where the MSX sounded like a NES.
one cant tell where is the limit, some samples from the amiga tracker samples disk may work, too.

technicaly it is like mp3 at low data rate.
the main point is that it can be used in a game.
not the "game over" sample at the end of a game, but in the middle of gaming action at tiny cpu cost.

By hit9918

Prophet (2868)

hit9918's picture

12-02-2017, 16:44

Quote:

I don't think it has an IFFT to return in time domain.

the soundchip is used to make those sinuses that are in an mp3.
in 50hz vblank interrupt rate the frequency registers and volume registers are updated.
an AY or an SCC used as mp3 decoder Smile
the AY has no sinuses but squares, another limitation.

By giangiacomo.zaffini

Champion (267)

giangiacomo.zaffini's picture

14-02-2017, 12:01

Ok, I see the explanation. Thank You.

Page 21/28
14 | 15 | 16 | 17 | 18 | 19 | 20 | | 22 | 23 | 24 | 25 | 26