Trixy/VoicePlay/main.py

#!/usr/bin/env python
# -*- coding: utf-8 -*-

import win32com.client
import platform
import wave
import numpy as np
from pydub import AudioSegment, silence
from pydub.playback import play
from playsound import playsound
import re
import librosa
from pydub import effects
import os

class Voice:

    
    instance = None
    @staticmethod
    def getInstance():
        if Voice.instance==None:
            Voice.instance = Voice()
        return Voice.instance
    
    def __init__(self):
        self.sys = platform.system()
        if self.sys=="Windows":
            self.speaker = win32com.client.Dispatch("SAPI.SpVoice")
        else:
            self.speaker = None

    def Say(self, text, save:bool = True):
        if self.speaker != None:
            self.speaker.Rate = 0.0
            if save:
                stream = win32com.client.Dispatch("SAPI.SpFileStream")
                stream.Open("cache/spoken.wav", 3, False)
                defAudioOutput = self.speaker.AudioOutputStream
                self.speaker.AudioOutputStream = stream
            self.speaker.Speak(text)
            if save:
                stream.Close()
                self.Edit("cache/spoken.wav","cache/pitched.wav")
                playsound("cache/pitched.wav")
                self.speaker.AudioOutputStream = defAudioOutput
        else:
            self.__svox(text)

    def EmotionalSay(self, text):
        files = list()
        defAudioOutput = self.speaker.AudioOutputStream
        collection = re.finditer('(?:\[(?P<args>.*?)\])?(?P<text>[^\[]*)', text, re.UNICODE)
        count=0
        combined_wav = AudioSegment.empty()
        self.speaker.Rate = -1
        for x in collection:
            args = self.__getArgs(x.group("args"))
            itm = self.__sayToFile(x.group("text"),"cache/tmp_"+str(count)+".wav")

            # Remove silence at the beginning/end
            start_trim = silence.detect_leading_silence(itm)
            end_trim = silence.detect_leading_silence(itm.reverse())
            duration = len(itm)
            itm = itm[start_trim:-end_trim]

            if args != "":
                itm_e = self.EditSegment(itm, args)
                combined_wav += itm_e
            else:
                combined_wav += itm
            #os.remove("cache/tmp_"+str(count)+".wav")
        self.speaker.AudioOutputStream = defAudioOutput
        play(combined_wav)

    def __sayToFile(self, text:str, file:str):
        if text == "":
            return AudioSegment.empty()

        stream = win32com.client.Dispatch("SAPI.SpFileStream")
        stream.Open(file, 3, False)
        self.speaker.AudioOutputStream = stream
        self.speaker.Speak(text)
        stream.Close()
        return AudioSegment.from_file(file)

    def PlaySound(file:str):
        if self.sys=="Windows":
            import winsound
            winsound.PlaySound("SystemExit", winsound.SND_ALIAS)

    def EditSegment(self, segment, args):
        print(args)
        if "vol" in args:
            segment += args["vol"]
        if "pitch" in args:
            segment = self.pitch_shift(segment, args["pitch"])
        if "speedup" in args:
            segment = segment.speedup(args["speedup"])
        if "speed" in args and args["speed"]>0:
            segment = self.speed_change(segment, 1.15)
            #segment = self.speed_change(segment, args["speed"])
            if args["speed"]>1:
                segment = self.pitch_shift(segment, -(1-args["speed"]))
            else:
                #pitch = (1/((args["speed"]/12)))*1
                pitch = 6.55
                #print("Speed Pitch: "+str(pitch))
                #segment = self.pitch_shift(segment, 1)
                #segment = segment.speedup(2)
        if "fadein" in args:
            segment = segment.fade_in(args["fadein"])
        if "fadeout" in args:
            segment = segment.fade_out(args["fadeout"])
        if "break" in args and args["break"]>=1:
            pauseSegment=AudioSegment.silent(args["break"])
            segment = pauseSegment + segment
        if "effect" in args:
            if os.path.exists("sounds/"+args["effect"]):
                fxSound=AudioSegment.from_file(args["effect"])
                segment = fxSound + segment
            else:
                print("  Effect not found")
        return segment

    def removeSilence(self, sound):
        non_sil_times = detect_nonsilent(sound, min_silence_len=50, silence_thresh=sound.dBFS * 1.5)
        if len(non_sil_times) > 0:
            non_sil_times_concat = [non_sil_times[0]]
            if len(non_sil_times) > 1:
                for t in non_sil_times[1:]:
                    if t[0] - non_sil_times_concat[-1][-1] < 200:
                        non_sil_times_concat[-1][-1] = t[1]
                    else:
                        non_sil_times_concat.append(t)
            non_sil_times = [t for t in non_sil_times_concat if t[1] - t[0] > 350]
            return sound[non_sil_times[0][0]: non_sil_times[-1][1]]

    def pitch_shift(self, sound, n_steps:float):
        y = np.frombuffer(sound._data, dtype=np.int16).astype(np.float32)/2**15
        y = librosa.effects.pitch_shift(y, sound.frame_rate, n_steps=n_steps)
        a  = AudioSegment(np.array(y * (1<<15), dtype=np.int16).tobytes(), frame_rate = sound.frame_rate, sample_width=2, channels = 1)
        return a

    def speed_change(self, sound, speed=1.0):
        # Manually override the frame_rate. This tells the computer how many
        # samples to play per second
        sound_with_altered_frame_rate = sound._spawn(sound.raw_data, overrides={
            "frame_rate": int(sound.frame_rate * speed)
        })
        # convert the sound with altered frame rate to a standard frame rate
        # so that regular playback programs will work right. They often only
        # know how to play audio at standard frame rate (like 44.1k)
        return sound_with_altered_frame_rate.set_frame_rate(sound.frame_rate)

    def detect_leading_silence(self, sound, silence_threshold=-50.0, chunk_size=10):
        '''
        sound is a pydub.AudioSegment
        silence_threshold in dB
        chunk_size in ms

        iterate over chunks until you find the first one with sound
        '''
        trim_ms = 0 # ms
        assert chunk_size > 0 # to avoid infinite loop
        while sound[trim_ms:trim_ms+chunk_size].dBFS < silence_threshold and trim_ms < len(sound):
            trim_ms += chunk_size
            return trim_ms

    def __getArgs(self, args:str):
        res = dict()
        if args is None or args=="":
            return res
        expl = args.split(",")
        for x in expl:
            x = x.strip()
            if ":" in x:
                k,v = x.split(":",2)
                print("getArgs -> '"+k+"' : '"+v+"'")
                if k.lower().strip()=="p" or k.lower().strip()=="pitch":
                    res["pitch"]=float(v)
                elif k.lower().strip()=="v" or k.lower().strip()=="vol" or k.lower().strip()=="volumen":
                    res["vol"]=int(v)
                elif k.lower().strip()=="su" or k.lower().strip()=="speedup":
                    res["speedup"]=float(v)
                elif k.lower().strip()=="s" or k.lower().strip()=="slow" or k.lower().strip()=="speed":
                    res["speed"]=float(v)
                elif k.lower().strip()=="b" or k.lower().strip()=="break" or k.lower().strip()=="pause" or k.lower().strip()=="p":
                    res["break"]=float(v)
                elif k.lower().strip()=="fade" or k.lower().strip()=="f":
                    res["fadein"]=int(v)
                    res["fadeout"]=int(v)
                elif k.lower().strip()=="fadein" or k.lower().strip()=="fi" or k.lower().strip()=="fade in" or k.lower().strip()=="fade-in":
                    res["fadein"]=int(v)
                elif k.lower().strip()=="fadeout" or k.lower().strip()=="fo" or k.lower().strip()=="fade out" or k.lower().strip()=="fade-out":
                    res["fadeout"]=int(v)
                elif k.lower().strip()=="fx" or k.lower().strip()=="effect":
                    res["effect"]=v
            else:
                if k.lower().strip()=="b" or k.lower().strip()=="break" or k.lower().strip()=="pause" or k.lower().strip()=="p":
                    res["break"]=1500
        return res

    def RobotEffect(self, input:str, output:str, frequence:int = 2):
        wr = wave.open(input, 'r')
        par = list(wr.getparams())
        par[3] = 0  # The number of samples will be set by writeframes.
        par = tuple(par)
        ww = wave.open(output, 'w')
        ww.setparams(par)
        sz = wr.getframerate()//frequence  # Read and process 1/fr second at a time.
        # A larger number for fr means less reverb.
        c = int(wr.getnframes()/sz)  # count of the whole file
        shift = 100//frequence  # shifting 100 Hz
        for num in range(c):
            da = np.fromstring(wr.readframes(sz), dtype=np.int16)
            left, right = da[0::2], da[1::2]  # left and right channel
            lf, rf = np.fft.rfft(left), np.fft.rfft(right)
            lf, rf = np.roll(lf, shift), np.roll(rf, shift)
            lf[0:shift], rf[0:shift] = 0, 0
            nl, nr = np.fft.irfft(lf), np.fft.irfft(rf)
            ns = np.column_stack((nl, nr)).ravel().astype(np.int16)
            ww.writeframes(ns.tostring())
        wr.close()
        ww.close()

    def __svox(self, text):
        pass