Organiza tus páginas con colecciones
Guarda y categoriza el contenido según tus preferencias.
En esta página, se incluyen recomendaciones para proporcionar datos de voz a la API de Asistente de Google. Estos lineamientos se diseñaron para aumentar la eficiencia y la exactitud, además de obtener tiempos de respuesta razonables del servicio.
Preprocesamiento de audio
Lo ideal es proporcionar un audio lo más limpio posible mediante un micrófono de buena calidad y correctamente ubicado. Sin embargo, la aplicación del procesamiento de la señal de reducción de ruidos al audio antes de enviarlo al servicio generalmente reduce la exactitud del reconocimiento. El servicio está diseñado para manejar audio ruidoso.
Para lograr resultados óptimos, haz lo siguiente:
Coloca el micrófono lo más cerca posible del usuario, en especial cuando hay ruido de fondo.
Evita el recorte de audio.
No uses el control automático de ganancia (AGC).
Todo el procesamiento de reducción de ruidos debe estar inhabilitado.
Idealmente, sigue estos pasos:
El nivel de audio debe calibrarse para que la señal de entrada no se recorte y los niveles máximos de audio de voz alcancen aproximadamente de -20 a -10 dBFS.
El dispositivo debe exhibir una amplitud "plana" frente a características de frecuencia (+- 3 dB, 100 Hz a 8,000 Hz).
La distorsión armónica total debe ser inferior al 1%, entre 100 Hz y 8,000 Hz, a un nivel de entrada de SPL de 90 dB.
[null,null,["Última actualización: 2025-07-26 (UTC)"],[[["\u003cp\u003eThe Google Assistant API is designed to handle noisy audio, so noise reduction processing should be disabled prior to sending audio.\u003c/p\u003e\n"],["\u003cp\u003eFor optimal performance, position the microphone near the user, avoid audio clipping and automatic gain control, and calibrate audio levels to peak between -20 to -10 dBFS.\u003c/p\u003e\n"],["\u003cp\u003eDevices used for audio input should ideally have a flat frequency response and minimal harmonic distortion.\u003c/p\u003e\n"],["\u003cp\u003e\u003cstrong\u003eWarning:\u003c/strong\u003e The Google Assistant Library for Python is deprecated; use the Google Assistant Service instead.\u003c/p\u003e\n"]]],["The Google Assistant Library for Python is deprecated; use the Google Assistant Service instead. For optimal audio input, provide clean audio from a high-quality, well-positioned microphone. Disable noise reduction and automatic gain control. Position the microphone close to the user and avoid audio clipping. Calibrate audio levels to avoid clipping, with peak levels around -20 to -10 dBFS. Aim for a flat amplitude response (+- 3 dB from 100 Hz to 8000 Hz) and total harmonic distortion under 1%.\n"],null,["# Best Practices for Audio\n\n| **Warning:** The Google Assistant Library for Python is deprecated as of June 28th, 2019. Use the [Google Assistant Service](/assistant/sdk/guides/service/python) instead.\n\n\u003cbr /\u003e\n\nThis page contains recommendations on how to provide speech data to the\nGoogle Assistant API. These guidelines are designed for greater efficiency\nand accuracy as well as reasonable response times from the service.\n\nAudio pre-processing\n--------------------\n\nIt's best to provide audio that is as clean as possible by using a good quality\nand well-positioned microphone. However, applying noise-reduction signal\nprocessing to the audio before sending it to the service typically reduces\nrecognition accuracy. The service is designed to handle noisy audio.\n\nFor best results:\n\n- Position the microphone as close to the user as possible, particularly when background noise is present.\n- Avoid audio clipping.\n- Do not use automatic gain control (AGC).\n- All noise reduction processing should be disabled.\n\nIdeally:\n\n- The audio level should be calibrated so that the input signal does not clip, and peak speech audio levels reach approximately -20 to -10 dBFS.\n- The device should exhibit approximately \"flat\" amplitude versus frequency characteristics (+- 3 dB 100 Hz to 8000 Hz).\n- Total harmonic distortion should be less than 1% from 100 Hz to 8000 Hz at 90 dB SPL input level."]]
