MasmBasic

[jj2007]

Not yet. I guess it's useless anyways because I have only one voice installed. Extra voices cost real money.

There is a demo here. Try with the text below - amazing!

There are many reasons why experienced programmers choose to write assembler code, performance issues where speed matters, the architectural freedom to lay out code in any way you like, the capacity to do things that cannot be done in many compilers but the main reason is simply because you can. Many conjure up the image of cobbling together a few DOS interrupts in unintelligible notation to prop up the shortcomings of compilers yet a modern assembler like MASM has the range of a high level language and can be written that way for high level code while retaining all of its power at the lowest level.

With the introduction of the 32 bit Windows API functions, MASM had access at the same functions that compilers had from the operating system but without the clutter and assumption of many of the compilers available. When you write Windows API code in MASM you get perfectly clear minimal precision code that leverages the full power of the Windows operating system and you get it at the code size you write, not with a pile of unwanted extras dumped into your executable by a compiler.

MASM has never been for the faint of heart, it is an uncompromised tool that has never been softened into a user friendly toy and it required the development of expertise to use correctly but for the programmer who already has experience in low level C and similar code, MASM offers power and flexibility that the best of compilers cannot deliver and contrary to popular opinion it can be developed and written at about the same development time as C code.

[Siekmanski]

I heard the text from your example. 
Copied it in the link you posted, there are some High Quality ones.

I have 2 amazingly High Quality SAPI5 Dutch female voices. (paid 200 euros for the 2 voices, today you can get 1 for 40 euros. )
Unfortunately they will only install on Windows 32bit. 

[jj2007]

I have downloaded the SDK from http://download.microsoft.com/download/B/4/3/B4314928-7B71-4336-9DE7-6FA4CF00B7B3/SpeechSDK51.exe

Mary & Mike don't work, and "Sample TTS voice" works but says most of the time "bla" (really )

[Siekmanski]

 
https://harposoftware.com/en/search?controller=search&orderby=position&orderway=desc&search_query=italian&submit_search=

[TimoVJL]

A test program for SAPI 5.x

Code: [Select]

#define WIN32_LEAN_AND_MEAN
#include <windows.h>
#include <sapi.h>

#pragma comment(lib, "ole32.lib")

int WINAPI WinMain(HINSTANCE hInstance, HINSTANCE hPrevInstance, LPSTR lpszCmdLine, int nCmdShow)
{
ISpVoice * pVoice = NULL;
CoInitialize(NULL);
HRESULT hr = CoCreateInstance(&CLSID_SpVoice, NULL, CLSCTX_ALL, &IID_ISpVoice, (void **)&pVoice);
if (SUCCEEDED(hr))
{
hr = pVoice->lpVtbl->Speak(pVoice, L"<voice optional='Gender=Male;'>Hello world", 0, NULL);
//hr = pVoice->lpVtbl->Speak(pVoice, L"<voice required='Gender = Male;'>Hello world", 0, NULL);
//if (hr) hr = pVoice->lpVtbl->Speak(pVoice, L"<voice required='Gender = Female;'>Hello world", 0, NULL);
pVoice->lpVtbl->Release(pVoice);
pVoice = NULL;
}
CoUninitialize();
return 0;
}

[jj2007]

Very interesting, Timo - I didn't know that you can specify options with XML tags:

Code: [Select]

hr = pVoice->lpVtbl->Speak(pVoice, L"<voice optional='Gender=Male;'>Hello world, I am a man", 0, NULL);
if (1)
hr = pVoice->lpVtbl->Speak(pVoice, L"<voice required='Gender=Female;'>Hello world, I am a woman", 0, NULL);With this modification, I hear both strings spoken by the same female voice when built as X64, but only "I am a woman" when built as x86 

With MasmBasic:

Say wChr$("<voice optional='Gender=lgbt;'>Good morning")


For Gender=, everything works fine except Gender=male, which is mute

[jj2007]

Update 28 July 2020 (in response to fast median thread and timings)

include \masm32\MasmBasic\MasmBasic.inc
  Init
  Dim MyArray() As REAL4        ; can be DWORD, REAL4 or REAL8
  For_ ecx=0 To 99
        Rand(-888.8, 999.9, MyArray(ecx))       ; put random numbers into the array
  Next
  PrintLine Str$("The elements:\t%i", MyArray(?))
  PrintLine Str$("The columns:\t%i", MyArray(?cols))    ; will return 0: this array is one-dimensional
  PrintLine Str$("The minimum:\t%f", MyArray(?min)v)    ; the functions marked with v return the
  PrintLine Str$("The maximum:\t%9f", MyArray(?max)v)   ; value in ST(0); the v tells Str$() to discard
  PrintLine Str$("The median:\t%f", MyArray(?median)v)  ; ST(0) with fstp st after printing
  PrintLine Str$("The mean:\t%9f", MyArray(?mean)v)     ; ?average does the same as ?mean
  PrintLine Str$("The sum: \t%9f", MyArray(?sum)v)      ; sum of all elements
EndOfCode

Code: [Select]

The elements:   100
The columns:    0
The minimum:    -884.9537
The maximum:    994.308716
The median:     -8.574001
The mean:       35.9553806
The sum:        3595.53806

[Siekmanski]

Cool, all in one. 

[guga]

Great work, JJ

One question about the median function.

Here

Code: [Select]

  fld st
  faddp st(2), st
  fsub arrInf.yMin ; make positive
  fmul arrInf.yRange ; normalise 0...99999
  ; fadd FP4(0.45)
  fistp arrInf.yIndex ; we need a
  mov eax, arrInf.yIndex ; reg32
  inc dword ptr [edi+4*eax] ; Populate the Median buffer with the values as index numbers and keep track for duplicates in one go
  dec ecx
  jns aiPop ; 0 is still ok

the size of the Table (MedBuffer) used in edi must be the double when we are dealing with negative values, right ?

I mean, say i have limits from -255 to 255, then the total size of the table should be (256*4*2) ?

Also, since it will be subtracting the negative values to set onto the proper address of the table, should it calculate the median and preserve the sign in case the resultant median is also negative ?

I previously normalized the inputed data to -1 to 1, so i divided each value with 255. On your version, at that routine, i made the following port:

Code: [Select]

; [Med_MinVal_Normalized: R$ -1.0] = < --- The minimum value as set in arrInf.yMin already normalized

    Do
        fld F$esi | fsub R$Med_MinVal_Normalized | fmul R$Float255 | fistp D@Index | mov eax D@Index ; Multiplied the input with 255 to it points to the proper address on the table.
        inc D$edi+eax*4   ; Populate the Median buffer with the values as index numbers
        add esi ebx       ; and keep track for duplicates in one go. And skip positions if needed
    Repeat_Until_Zero ecx

Is this correct ? Will it result a negative median (thus, preserving the sign) if the median is, in fact, negative ?

[jj2007]

Hi Guga,
In the example above the range of values is determined by the Rand(-888.8, 999.9, destination). The normalisation takes care of negative numbers. In fact, the median for this dataset is slightly negative.

Code: [Select]

The minimum:    -884.9537
The maximum:    994.308716
The median:     -8.574001
The mean:       35.9553806

[guga]

Hi JJ, ok, but how the normalization works ? The math operation, i mean.

I saw it takes the subtraction of the Maxvalue to the MinValue, but it will divide by what ?

Fraction = (Max-Min)/ WhatData ?

What Data is suppose to be divided with ? If we have negative numbers, it means we have double the amount of values, right ? ex, if the limits to be settled are -255 to 255, it means we have, in fact a limit of 255*2 data to load. So, if our Minimum value found is, let´s say -200 and Max is 185, then to apply the normalization, is correct to do this ?

Fraction = (185-(-200))(2*255) = (185+200)/510 = 385/510 = 0.754901961

So, our Fraction (arrInf.yRange) would be 0.754901961, right ?

And to recover the proper median at the end of the function, we do this ?

New Value = ValueFound/Fraction - Minimum Value ? = ValueFound/0.75490191 - (-200) ?

[jj2007]

Hi JJ, ok, but how the normalization works ? The math operation, i mean.

Search for arrInf.yRange - it should be self-explanatory