Spaces:
Runtime error
Runtime error
File size: 3,545 Bytes
76b7899 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 |
import gradio as gr
import librosa
import torch
from transformers import AutoTokenizer, pipeline, logging
from auto_gptq import AutoGPTQForCausalLM, BaseQuantizeConfig
from transformers import SpeechT5Processor, SpeechT5ForSpeechToText
# Audio
checkpoint = "microsoft/speecht5_asr"
audio_processor = SpeechT5Processor.from_pretrained(checkpoint)
audio_model = SpeechT5ForSpeechToText.from_pretrained(checkpoint)
def process_audio(sampling_rate, waveform):
# convert from int16 to floating point
waveform = waveform / 32678.0
# convert to mono if the stereo
if len(waveform.shape) > 1:
waveform = librosa.to_mono(waveform.T)
# resample to 16 kHz if necessary
if sampling_rate != 16000:
waveform = librosa.resample(waveform, orig_sr=sampling_rate, target_sr=16000)
# limit to 30 seconds
waveform = waveform[:16000*30]
# make PyTorch tensor
waveform = torch.tensor(waveform)
return waveform
def audio_to_text(audio, mic_audio=None):
# audio = tuple (sample_rate, frames) or (sample_rate, (frames, channels))
if mic_audio is not None:
sampling_rate, waveform = mic_audio
elif audio is not None:
sampling_rate, waveform = audio
else:
return "(please provide audio)"
waveform = process_audio(sampling_rate, waveform)
inputs = audio_processor(audio=waveform, sampling_rate=16000, return_tensors="pt")
predicted_ids = audio_model.generate(**inputs, max_length=400)
transcription = audio_processor.batch_decode(predicted_ids, skip_special_tokens=True)
return transcription[0]
# Text Generation
model_name_or_path = "TheBloke/Llama-2-7b-Chat-GPTQ"
model_basename = "gptq_model-4bit-128g"
use_triton = False
llama_tokenizer = AutoTokenizer.from_pretrained(model_name_or_path, use_fast=True)
llama_model = AutoGPTQForCausalLM.from_quantized(model_name_or_path,
model_basename=model_basename,
use_safetensors=True,
trust_remote_code=True,
device="cuda:0",
use_triton=use_triton,
quantize_config=None)
def generate(text):
prompt = text
system_message = "You are a helpful, respectful and honest assistant. Always answer as helpfully as possible, while being safe. Your answers should not include any harmful, unethical, racist, sexist, toxic, dangerous, or illegal content. Please ensure that your responses are socially unbiased and positive in nature. If a question does not make any sense, or is not factually coherent, explain why instead of answering something not correct. If you don't know the answer to a question, please don't share false information. Give short, simple and direct answers"
prompt_template=f'''[INST] <<SYS>>
{system_message}
<</SYS>>
{prompt} [/INST]'''
pipe = pipeline(
"text-generation",
model=model,
tokenizer=tokenizer,
max_new_tokens=512,
temperature=0.7,
top_p=0.95,
repetition_penalty=1.15
)
return pipe(prompt_template)[0]['generated_text']
def audio_text_generate(audio):
audio_text = audio_to_text(audio)
generated_text = generate(audio_text)
response = generated_text[generated_text.index("[/INST]")+7:].strip()
return audio_text, response
demo = gr.Interface(fn=audio_text_generate,
inputs=gr.Audio(source="microphone"),
outputs=[gr.Text(label="Audio Text"), gr.Text(label="Generated Text")])
# examples=["https://samplelib.com/lib/preview/mp3/sample-3s.mp3"], cache_examples=True)
demo.launch() |