I want to transcribe audio in real-time concurrently with recording using PyAudio's stream

[houkagoplay]

"I'm currently developing a program using PyAudio to record microphone audio and perform real-time speech-to-text transcription. I've successfully saved the audio data as a file using stream.read, but I'm struggling to perform the transcription. Instead of transcribing the saved file immediately after recording, I want to transcribe it concurrently with the recording process. However, I'm unsure how to pass the audio data from the streaming process to the transcription process.

In the example provided by the library:

segments, info = model.transcribe("audio.mp3", beam_size=5)

Is it possible to use the audio data captured during recording in place of "audio.mp3"? If so, what processing steps are needed to pass the streaming data for transcription?

I would appreciate any guidance or suggestions on how to achieve this.

This document was created using Google Translate and chatgpt.

[MahmoudAshraf97]

You can pass a float32 array sampled at 16khz

[houkagoplay]

Thank you!! I will try that!!

[trungkienbkhn]

You can try using the 'sounddevice' lib:

import numpy as np
import sounddevice as sd

from faster_whisper import WhisperModel

print("Recording started")
duration = 10
sample_rate = 16000
audio_data = sd.rec(
    int(sample_rate * duration), samplerate=sample_rate, channels=1, dtype=np.float32
)
sd.wait()
audio_data = audio_data.squeeze()
print(audio_data)
print("Recording stopped")

model = WhisperModel("tiny", device="cpu")
segments, info = model.transcribe(audio_data, word_timestamps=True)
for segment in segments:
    print("[%.2fs -> %.2fs] %s" % (segment.start, segment.end, segment.text))

[houkagoplay]

Thank you! I'll try this method too!