File size: 4,095 Bytes
efc446d
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
import whisper
from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.metrics.pairwise import cosine_similarity
from pydub import AudioSegment
from hezar.models import Model
import librosa
import soundfile as sf
from audio_separator.separator import Separator
from logging import ERROR
import streamlit as st


def cosine_sim(text1, text2):
    vectorizer = TfidfVectorizer().fit_transform([text1, text2])
    vectors = vectorizer.toarray()
    return cosine_similarity(vectors)[0, 1]
    
def take_challenge(music_file, typed_lyrics, key, language, has_background=False, background_audio_path=None):
    st.write("Listen to music since you have to record 15seconds after that")
    st.audio(music_file)
    if has_background:
        st.write("Play this music while singing which might help you:")
        st.audio(background_audio_path)
    audio_value = st.experimental_audio_input("Sing Rest of music:🎙️", key=key)
    if audio_value:
        with open("user_sing.mp3", "wb") as f:
            f.write(audio_value.getbuffer())
        
        if has_background:
            file_to_transcribe = split_vocals("user_sing.mp3")[1]
        else:
            file_to_transcribe = "user_sing.mp3"


        if language == "en":
            english_model = whisper.load_model("base.en")
            user_lyrics = english_model.transcribe(file_to_transcribe, language=language)["text"]
        else:
            persian_model = Model.load("hezarai/whisper-small-fa")
            user_lyrics = persian_model.predict(file_to_transcribe)[0]["text"]
            
        st.write(user_lyrics)
        similarity_score = cosine_sim(typed_lyrics, user_lyrics)
        if similarity_score > 0.85:
            st.success('Awsome! You are doing great', icon="✅")
            st.markdown('<style>div.stAlert { background-color: rgba(3, 67, 24, 0.9); }</style>', unsafe_allow_html=True)
        else:
            st.error('Awful! Try harder next time', icon="🚨")
            st.markdown('<style>div.stAlert { background-color: rgba(241, 36, 36, 0.9); }</style>', unsafe_allow_html=True)

def change_volume(input_file, output_file, volume_factor):
    sound = AudioSegment.from_mp3(input_file)
    volume_changed = sound + volume_factor
    volume_changed.export(output_file, format="mp3")

def change_speed(input_file, output_file, speed_factor):
    sound, sr = librosa.load(input_file)
    speed_changed = librosa.effects.time_stretch(sound, rate=speed_factor)
    sf.write(output_file, speed_changed, sr)

def change_pitch(input_file, output_file, pitch_factor):
    sound, sr = librosa.load(input_file)
    pitch_changed = librosa.effects.pitch_shift(sound, sr=sr, n_steps=pitch_factor)
    sf.write(output_file, pitch_changed, sr)

def low_pass_filter(input_file, output_file, cutoff_freq):
    sound = AudioSegment.from_mp3(input_file)
    low_filtered_sound = sound.low_pass_filter(cutoff_freq)
    low_filtered_sound.export(output_file, format="mp3")

def high_pass_filter(input_file, output_file, cutoff_freq):
    sound = AudioSegment.from_mp3(input_file)
    high_filtered_sound = sound.high_pass_filter(cutoff_freq)
    high_filtered_sound.export(output_file, format="mp3")

def pan_left_right(input_file, output_file, pan_factor):
    sound = AudioSegment.from_mp3(input_file)
    pan_sound = sound.pan(pan_factor)
    pan_sound.export(output_file, format="mp3")

def fade_in_ms(input_file, output_file, fade_factor):
    sound = AudioSegment.from_mp3(input_file)
    faded_sound = sound.fade_in(fade_factor)
    faded_sound.export(output_file, format="mp3")

def fade_out_ms(input_file, output_file, fade_factor):
    sound = AudioSegment.from_mp3(input_file)
    faded_sound = sound.fade_out(fade_factor)
    faded_sound.export(output_file, format="mp3")

def split_vocals(input_file):
    separator = Separator(output_format="mp3", log_level=ERROR)
    separator.load_model("MGM_MAIN_v4.pth")
    result_list = separator.separate(input_file, primary_output_name=input_file[:-4]+"_instruments", secondary_output_name=input_file[:-4]+"_vocals")
    return result_list