altera o script para usar pywin32 em vez de win32gui, dá erro, e imprime o resultado,…
- User
altera o script para usar pywin32 em vez de win32gui, dá erro, e imprime o resultado, além da janela windows, também no prompt:
import whisper import pyaudio import wave import win32gui import time import threading import keyboard
model = whisper.load_model("small")
def transcribe(audio):
# load audio and pad/trim it to fit 30 seconds audio = whisper.load_audio(audio) audio = whisper.pad_or_trim(audio) # make log-Mel spectrogram and move to the same device as the model mel = whisper.log_mel_spectrogram(audio).to(model.device) # detect the spoken language _, probs = model.detect_language(mel) print(f"Detected language: {max(probs, key=probs.get)}") # decode the audio options = whisper.DecodingOptions(fp16 = False) result = whisper.decode(model, mel, options) return result.textCHUNK = 1024 FORMAT = pyaudio.paInt16 # Audio format CHANNELS = 1 # Number of channels RATE = 44100 # Sampling rate (number of samples per second) THRESHOLD = 500
p = pyaudio.PyAudio()
stream = p.open(format=FORMAT, channels=CHANNELS, rate=RATE, input=True, frames_per_buffer=CHUNK)
frames = [] recording = False
print("Waiting for voice...")
while True: data = stream.read(CHUNK) rms = int.from_bytes(data, byteorder='little')
if rms > THRESHOLD and not recording: print("Recording...") recording = True if recording: frames.append(data) if len(frames) > RATE / CHUNK * 5: breakprint("Recording finished!")
stream.stop_stream() stream.close() p.terminate()
wf = wave.open("output.wav", "wb") wf.setnchannels(CHANNELS) wf.setsampwidth(p.get_sample_size(FORMAT)) wf.setframerate(RATE) wf.writeframes(b"".join(frames)) wf.close()
def write_in_window(text): # get the handle of the current window handle = win32gui.GetForegroundWindow() # write text in the window win32gui.SendMessage(handle, 0x000C, 0, text)
def transcribe_thread(): while True: transcription = transcribe("output.wav") print(transcription) # print transcription to console write_in_window(transcription) time.sleep(5)
t1 = threading.Thread(target=transcribe_thread) t1.start()
while True: if keyboard.is_pressed('ctrl+end'): break elif keyboard.is_pressed('ctrl+home'): if t1.is_alive(): t1.join() else: t1.start() Here’s an example of how you could modify the transcribe_thread function to continuously record and transcribe new audio:
def transcribe_thread(): while True: # Record new audio frames = [] recording = False print("Waiting for voice...") while True: data = stream.read(CHUNK) rms = int.from_bytes(data, byteorder='little') if rms > THRESHOLD and not recording: print("Recording...") recording = True if recording: frames.append(data) if len(frames) > RATE / CHUNK * 5: break print("Recording finished!") # Save recorded audio to file wf = wave.open("output.wav", "wb") wf.setnchannels(CHANNELS) wf.setsampwidth(p.get_sample_size(FORMAT)) wf.setframerate(RATE) wf.writeframes(b"".join(frames)) wf.close() # Transcribe recorded audio transcription = transcribe("output.wav") print(transcription) # print transcription to console write_in_window(transcription)
adapta o codigo anterior a esta alteração
- Assistant
import whisper import pyaudio import wave import win32com.client import time import threading import keyboard
model = whisper.load_model("small")
def transcribe(audio):
# load audio and pad/trim it to fit 30 seconds audio = whisper.load_audio(audio) audio = whisper.pad_or_trim(audio) # make log-Mel spectrogram and move to the same device as the model mel = whisper.log_mel_spectrogram(audio).to(model.device) # detect the spoken language _, probs = model.detect_language(mel) print(f"Detected language: {max(probs, key=probs.get)}") # decode the audio options = whisper.DecodingOptions(fp16 = False) result = whisper.decode(model, mel, options) return result.textCHUNK = 1024 FORMAT = pyaudio.paInt16 # Audio format CHANNELS = 1 # Number of channels RATE = 44100 # Sampling rate (number of samples per second) THRESHOLD = 500
p = pyaudio.PyAudio()
stream = p.open(format=FORMAT, channels=CHANNELS, rate=RATE, input=True, frames_per_buffer=CHUNK)
frames = [] recording = False
print("Waiting for voice...")
while True: data = stream.read(CHUNK) rms = int.from_bytes(data, byteorder='little')
if rms > THRESHOLD and not recording: print("Recording...") recording = True if recording: frames.append(data) if len(frames) > RATE / CHUNK * 5: breakprint("Recording finished!")
stream.stop_stream() stream.close() p.terminate()
wf = wave.open("output.wav", "wb") wf.setnchannels(CHANNELS) wf.setsampwidth(p.get_sample_size(FORMAT)) wf.setframerate(RATE) wf.writeframes(b"".join(frames)) wf.close()
def write_in_window(text): # get the handle of the current window pythoncom.CoInitialize() # initialize pywin32 handle = win32com.client.Dispatch("WScript.Shell").AppActivate("Untitled - Notepad") # replace "Untitled - Notepad" with the title of your window win32com.client.Dispatch("WScript.Shell").SendKeys(text) pythoncom.CoUninitialize()
def transcribe_thread(): while True: # Record new audio frames = [] recording = False print("Waiting for voice...") while True: data = stream.read(CHUNK) rms = int.from_bytes(data, byteorder='little') if rms > THRESHOLD and not recording: print("Recording...") recording = True if recording: frames.append(data) if len(frames) > RATE / CHUNK * 5: break print("Recording finished!") # Save recorded audio to file wf = wave.open("output.wav", "wb") wf.setnchannels(CHANNELS) wf.setsampwidth(p.get_sample_size(FORMAT)) wf.setframerate(RATE) wf.writeframes(b"".join(frames)) wf.close() # Transcribe recorded audio transcription = transcribe("output.wav") print(transcription) # print transcription to console write_in_window(transcription) print(transcription) # print transcription to console
t1 = threading.Thread(target=transcribe_thread) t1.daemon = True t1.start()
while True: if keyboard.is_pressed('ctrl+end'): break elif keyboard.is_pressed('ctrl+home'): if t1.is_alive(): t1.join() else: t1.start()
- User
faz identações no codigo
- Assistant
import whisper import pyaudio import wave import win32com.client import time import threading import keyboard
model = whisper.load_model("small")
def transcribe(audio):
# load audio and pad/trim it to fit 30 seconds audio = whisper.load_audio(audio) audio = whisper.pad_or_trim(audio) # make log-Mel spectrogram and move to the same device as the model mel = whisper.log_mel_spectrogram(audio).to(model.device) # detect the spoken language _, probs = model.detect_language(mel) print(f"Detected language: {max(probs, key=probs.get)}") # decode the audio options = whisper.DecodingOptions(fp16 = False) result = whisper.decode(model, mel, options) return result.textCHUNK = 1024 FORMAT = pyaudio.paInt16 # Audio format CHANNELS = 1 # Number of channels RATE = 44100 # Sampling rate (number of samples per second) THRESHOLD = 500
p = pyaudio.PyAudio()
stream = p.open(format=FORMAT, channels=CHANNELS, rate=RATE, input=True, frames_per_buffer=CHUNK)
frames = [] recording = False
print("Waiting for voice...")
while True: data = stream.read(CHUNK) rms = int.from_bytes(data, byteorder='little')
if rms > THRESHOLD and not recording: print("Recording...") recording = True if recording: frames.append(data) if len(frames) > RATE / CHUNK * 5: breakprint("Recording finished!")
stream.stop_stream() stream.close() p.terminate()
wf = wave.open("output.wav", "wb") wf.setnchannels(CHANNELS) wf.setsampwidth(p.get_sample_size(FORMAT)) wf.setframerate(RATE) wf.writeframes(b"".join(frames)) wf.close()
def write_in_window(text): # get the handle of the current window pythoncom.CoInitialize() # initialize pywin32 handle = win32com.client.Dispatch("WScript.Shell").AppActivate("Untitled - Notepad") # replace "Untitled - Notepad" with the title of your window win32com.client.Dispatch("WScript.Shell").SendKeys(text) pythoncom.CoUninitialize()
def transcribe_thread(): while True: # Record new audio frames = [] recording = False print("Waiting for voice...") while True: data = stream.read(CHUNK) rms = int.from_bytes(data, byteorder='little') if rms > THRESHOLD and not recording: print("Recording...") recording = True if recording: frames.append(data) if len(frames) > RATE / CHUNK * 5: break print("Recording finished!") # Save recorded audio to file wf = wave.open("output.wav", "wb") wf.setnchannels(CHANNELS) wf.setsampwidth(p.get_sample_size(FORMAT)) wf.setframerate(RATE) wf.writeframes(b"".join(frames)) wf.close() # Transcribe recorded audio transcription = transcribe("output.wav") print(transcription) # print transcription to console write_in_window(transcription) print(transcription) # print transcription to console
t1 = threading.Thread(target=transcribe_thread) t1.daemon = True t1.start()
while True: if keyboard.is_pressed('ctrl+end'): break elif keyboard.is_pressed('ctrl+home'): if t1.is_alive(): t1.join() else: t1.start()
- User
erro: pip install pywintypes Defaulting to user installation because normal site-packages is not writeable ERROR: Could not find a version that satisfies the requirement pywintypes (from versions: none) ERROR: No matching distribution found for pywintypes
- Assistant
Desculpe, foi um engano meu.
pywintypesé uma dependência dopywin32, que você já tem instalado.O erro possivelmente está ocorrendo porque, ao instalar o
pywin32, você selecionou a opção "Executable installers", que é recomendada somente se você estiver usando o Python 2.7 ou o Python 3.3. Para outras versões do Python, é recomendado que se use a opção "Wheel installers", que é mais moderna.Para corrigir o erro, sugiro que você desinstale o
pywin32e calcule a opção "Wheel installers" ao reinstalá-lo. Para tal, execute os seguintes comandos:pip uninstall pywin32 pip install pywin32 --no-binary pywin32Isso deve resolver o erro.
- User
pip install pywin32 --no-binary pywin32 Defaulting to user installation because normal site-packages is not writeable ERROR: Could not find a version that satisfies the requirement pywin32 (from versions: none) ERROR: No matching distribution found for pywin32
- Assistant
Nesse caso, você pode tentar instalar o
pywin32manualmente. Acesse o site https://sourceforge.net/projects/pywin32/files/pywin32/ e baixe o arquivo correspondente à sua versão do Python e à arquitetura do seu sistema operacional.Depois de fazer o download, extraia o arquivo e execute o instalador
pywin32-xxx.winxx-pyxx.exe. Certifique-se de selecionar a opção "Complete" (instala todos os módulos) e aguarde até que a instalação seja concluída.Caso o erro persista após a instalação manual, verifique se o seu ambiente virtual Python está ativado e se você está usando a última versão do pip. Se ainda houver problema, entre em contato novamente, por favor.
- User
mostra o codigo python linha a linha, devidamente indentado: python # import whisper # import pyaudio # import wave # import win32gui # import time # import threading # import keyboard # import win32con # import win32api model = whisper.load_model("small") def transcribe(audio): # Carregar áudio e ajustar para 30 segundos audio = whisper.load_audio(audio) audio = whisper.pad_or_trim(audio) # Gerar espectrograma Mel-log e mover para o mesmo dispositivo do modelo mel = whisper.log_mel_spectrogram(audio).to(model.device) # Detectar o idioma falado _, probs = model.detect_language(mel) print(f"Idioma detectado: {max(probs, key=probs.get)}") # Decodificar o áudio options = whisper.DecodingOptions(fp16=False) result = whisper.decode(model, mel, options) return result.text CHUNK = 1024 FORMAT = pyaudio.paInt16 CHANNELS = 1 RATE = 44100 THRESHOLD = 500 p = pyaudio.PyAudio() stream = p.open(format=FORMAT, channels=CHANNELS, rate=RATE, input=True, frames_per_buffer=CHUNK) frames = [] recording = False print("Aguardando voz...") while True: data = stream.read(CHUNK) rms = int.from_bytes(data, byteorder='little') if rms > THRESHOLD and not recording: print("Gravando...") recording = True if recording: frames.append(data) if len(frames) > RATE / CHUNK * 5: break print("Gravação finalizada!") stream.stop_stream() stream.close() p.terminate() wf = wave.open("output.wav", "wb") wf.setnchannels(CHANNELS) wf.setsampwidth(p.get_sample_size(FORMAT)) wf.setframerate(RATE) wf.writeframes(b"".join(frames)) wf.close() # def write_in_window(text): # while True: # try: # handle = win32gui.GetForegroundWindow() # win32api.SendMessage(handle, win32con.WM_SETTEXT, 0, text) # break # except: # continue # def transcribe_thread(): # while True: # transcription = transcribe("output.wav") # print(transcription) # imprimir a transcrição no console # write_in_window(transcription) # time.sleep(5) # t1 = threading.Thread(target=transcribe_thread) # t1.start() # while True: # if keyboard.is_pressed('ctrl+end'): # break # elif keyboard.is_pressed('ctrl+home'): # if t1.is_alive(): # t1.join() # else: # t1.start()
- Assistant
import whisper import pyaudio import wave import win32gui import time import threading import keyboard import win32con import win32api model = whisper.load_model("small") def transcribe(audio): # Carregar áudio e ajustar para 30 segundos audio = whisper.load_audio(audio) audio = whisper.pad_or_trim(audio) # Gerar espectrograma Mel-log e mover para o mesmo dispositivo do modelo mel = whisper.log_mel_spectrogram(audio).to(model.device) # Detectar o idioma falado _, probs = model.detect_language(mel) print(f"Idioma detectado: {max(probs, key=probs.get)}") # Decodificar o áudio options = whisper.DecodingOptions(fp16=False) result = whisper.decode(model, mel, options) return result.text CHUNK = 1024 FORMAT = pyaudio.paInt16 CHANNELS = 1 RATE = 44100 THRESHOLD = 500 p = pyaudio.PyAudio() stream = p.open(format=FORMAT, channels=CHANNELS, rate=RATE, input=True, frames_per_buffer=CHUNK) frames = [] recording = False print("Aguardando voz...") while True: data = stream.read(CHUNK) rms = int.from_bytes(data, byteorder='little') if rms > THRESHOLD and not recording: print("Gravando...") recording = True if recording: frames.append(data) if len(frames) > RATE / CHUNK * 5: break print("Gravação finalizada!") stream.stop_stream() stream.close() p.terminate() wf = wave.open("output.wav", "wb") wf.setnchannels(CHANNELS) wf.setsampwidth(p.get_sample_size(FORMAT)) wf.setframerate(RATE) wf.writeframes(b"".join(frames)) wf.close() # def write_in_window(text): # while True: # try: # handle = win32gui.GetForegroundWindow() # win32api.SendMessage(handle, win32con.WM_SETTEXT, 0, text) # break # except: # continue # def transcribe_thread(): # while True: # transcription = transcribe("output.wav") # print(transcription) # imprimir a transcrição no console # write_in_window(transcription) # time.sleep(5) # t1 = threading.Thread(target=transcribe_thread) # t1.start() # while True: # if keyboard.is_pressed('ctrl+end'): # break # elif keyboard.is_pressed('ctrl+home'): # if t1.is_alive(): # t
Other models, same or similar prompt
Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.