ctcalign.py

from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor
import torch, torchaudio
import soundfile as sf
import numpy as np
import scipy.signal

#------------------------------------------
# setup wav2vec2
#------------------------------------------

device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
torch.random.manual_seed(0)

# info: https://huggingface.co/carlosdanielhernandezmena/wav2vec2-large-xlsr-53-icelandic-ep10-1000h/blob/main/vocab.json
is_MODEL_PATH="carlosdanielhernandezmena/wav2vec2-large-xlsr-53-icelandic-ep10-1000h"
is_model_blank_token = '[PAD]' # important to know for CTC decoding
is_model_word_separator = '|'
is_labels_dict = {"f": 0, "a": 1, "é": 2, "t": 3, "o": 4, "n": 5, "e": 6, "y": 8, "k": 9, "j": 10, "u": 11, "d": 12, "w": 13, "l": 14, "ú": 15, "q": 16, "g": 17, "í": 18, "s": 19, "r": 20, "ý": 21, "i": 22, "z": 23, "m": 24, "h": 25, "ó": 26, "þ": 27, "æ": 28, "c": 29, "á": 30, "v": 31, "b": 32, "ð": 33, "x": 34, "ö": 35, "p": 36, "|": 7, "[UNK]": 37, "[PAD]": 38}

is_model = Wav2Vec2ForCTC.from_pretrained(is_MODEL_PATH).to(device)
is_processor = Wav2Vec2Processor.from_pretrained(is_MODEL_PATH)
is_inverse_dict = {v:k for k,v in is_labels_dict.items()}
is_all_labels = tuple(is_labels_dict.keys())
is_blank_id = is_labels_dict[is_model_blank_token]


fo_MODEL_PATH="carlosdanielhernandezmena/wav2vec2-large-xlsr-53-faroese-100h"
fo_model_blank_token = '[PAD]' # important to know for CTC decoding
fo_model_word_separator = '|'
fo_labels_dict = {"w": 0, "i": 1, "6": 2, "s": 3, "_": 4, "k": 5, "l": 6, "ú": 7, "2": 8, "4": 9, "d": 10, "z": 11, "3": 12, "ð": 13, "t": 15, "ø": 16, "x": 17, "p": 18, "o": 19, "æ": 20, "n": 21, "f": 22, "á": 23, "5": 24, "g": 25, "ý": 26, "r": 27, "é": 28, "u": 29, "ü": 30, "y": 31, "í": 32, "h": 33, "q": 34, "b": 35, "e": 36, "v": 37, "-": 38, "c": 39, "j": 40, ".": 41, "ó": 42, "'": 43, "m": 44, "a": 45, "|": 14, "[UNK]": 46, "[PAD]": 47}

fo_model = Wav2Vec2ForCTC.from_pretrained(fo_MODEL_PATH).to(device)
fo_processor = Wav2Vec2Processor.from_pretrained(fo_MODEL_PATH)
fo_inverse_dict = {v:k for k,v in fo_labels_dict.items()}
fo_all_labels = tuple(fo_labels_dict.keys())
fo_blank_id = fo_labels_dict[fo_model_blank_token]


no_MODEL_PATH="NbAiLab/nb-wav2vec2-1b-bokmaal"
no_model_blank_token = '[PAD]' # important to know for CTC decoding
no_model_word_separator = '|'
no_labels_dict = {"a": 1, "b": 2, "c": 3, "d": 4, "e": 5, "f": 6, "g": 7, "h": 8, "i": 9, "j": 10, "k": 11, "l": 12, "m": 13, "n": 14, "o": 15, "p": 16, "q": 17, "r": 18, "s": 19, "t": 20, "u": 21, "v": 22, "w": 23, "x": 24, "y": 25, "z": 26, "å": 27, "æ": 28, "ø": 29, "|": 0, "[UNK]": 30, "[PAD]": 31}

no_model = Wav2Vec2ForCTC.from_pretrained(no_MODEL_PATH).to(device)
no_processor = Wav2Vec2Processor.from_pretrained(no_MODEL_PATH)
no_inverse_dict = {v:k for k,v in no_labels_dict.items()}
no_all_labels = tuple(no_labels_dict.keys())
no_blank_id = no_labels_dict[no_model_blank_token]


d = {"Icelandic": {'model': is_model, 'processor': is_processor, 'inverse_dict': is_inverse_dict, 'labels_dict': is_labels_dict, 'all_labels': is_all_labels, 'blank_id': is_blank_id, 'model_blank_token': is_model_blank_token, 'model_word_separator': is_model_word_separator}, "Faroese": {'model': fo_model, 'processor': fo_processor, 'inverse_dict': fo_inverse_dict, 'labels_dict': fo_labels_dict, 'all_labels': fo_all_labels, 'blank_id': fo_blank_id, 'model_blank_token': fo_model_blank_token, 'model_word_separator': fo_model_word_separator}, "Norwegian": {'model': no_model, 'processor': no_processor, 'inverse_dict': no_inverse_dict, 'labels_dict': no_labels_dict, 'all_labels': no_all_labels, 'blank_id': no_blank_id, 'model_blank_token': no_model_blank_token, 'model_word_separator': no_model_word_separator} }








#------------------------------------------
# forced alignment with ctc decoder
#   originally based on implementation of
#   https://pytorch.org/audio/main/tutorials/forced_alignment_tutorial.html
#------------------------------------------

# return the label class probability of each audio frame
def get_frame_probs(wav_path,lang):
    wav = readwav(wav_path)
    with torch.inference_mode(): # similar to with torch.no_grad():
        input_values = d[lang]['processor'](wav,sampling_rate=16000).input_values[0]
        input_values = torch.tensor(input_values, device=device).unsqueeze(0)
        emits = d[lang]['model'](input_values).logits
        emits = torch.log_softmax(emits, dim=-1)
        emit = emits[0].cpu().detach()
    return emit


def get_trellis(emission, tokens, blank_id):
    num_frame = emission.size(0)
    num_tokens = len(tokens)
    # Trellis has extra diemsions for both time axis and tokens.
    # The extra dim for tokens represents <SoS> (start-of-sentence)
    # The extra dim for time axis is for simplification of the code.
    trellis = torch.empty((num_frame + 1, num_tokens + 1))
    trellis[0, 0] = 0
    trellis[1:, 0] = torch.cumsum(emission[:, 0], 0) # len of this slice of trellis is len of audio frames)
    trellis[0, -num_tokens:] = -float("inf") # len of this slice of trellis is len of transcript tokens
    trellis[-num_tokens:, 0] = float("inf")
    for t in range(num_frame):
        trellis[t + 1, 1:] = torch.max(
            # Score for staying at the same token
            trellis[t, 1:] + emission[t, blank_id],
            # Score for changing to the next token
            trellis[t, :-1] + emission[t, tokens],
        )
    return trellis


def backtrack(trellis, emission, tokens, blank_id):
    # Note:
    # j and t are indices for trellis, which has extra dimensions
    # for time and tokens at the beginning.
    # When referring to time frame index `T` in trellis,
    # the corresponding index in emission is `T-1`.
    # Similarly, when referring to token index `J` in trellis,
    # the corresponding index in transcript is `J-1`.
    j = trellis.size(1) - 1
    t_start = torch.argmax(trellis[:, j]).item()

    path = []
    for t in range(t_start, 0, -1):
        # 1. Figure out if the current position was stay or change
        # Note (again):
        # `emission[J-1]` is the emission at time frame `J` of trellis dimension.
        # Score for token staying the same from time frame J-1 to T.
        stayed = trellis[t - 1, j] + emission[t - 1, blank_id]
        # Score for token changing from C-1 at T-1 to J at T.
        changed = trellis[t - 1, j - 1] + emission[t - 1, tokens[j - 1]]

        # 2. Store the path with frame-wise probability.
        prob = emission[t - 1, tokens[j - 1] if changed > stayed else 0].exp().item()
        # Return token index and time index in non-trellis coordinate.
        path.append((j - 1, t - 1, prob))

        # 3. Update the token
        if changed > stayed:
            j -= 1
            if j == 0:
                break
    else:
        raise ValueError("Failed to align")
    return path[::-1]



def merge_repeats(path,transcript):
    i1, i2 = 0, 0
    segments = []
    while i1 < len(path):
        while i2 < len(path) and path[i1][0] == path[i2][0]: # while both path steps point to the same token index
            i2 += 1
        segments.append( # when i2 finally switches to a different token,
            #Segment(
                (transcript[path[i1][0]], # to the list of segments, append the token from i1 
                path[i1][1], # time of the first path-point of that token
                path[i2 - 1][1] + 1, # time of the final path-point for that token.
            )
        )
        i1 = i2
    return segments

def merge_words(segments, separator):
    words = []
    i1, i2 = 0, 0
    while i1 < len(segments):
        if i2 >= len(segments) or segments[i2][0] == separator:
            if i1 != i2:
                segs = segments[i1:i2]
                word = "".join([seg[0] for seg in segs])
                words.append((word, segments[i1][1], segments[i2 - 1][2]))
            i1 = i2 + 1
            i2 = i1
        else:
            i2 += 1
    return words




#------------------------------------------
# handle in/out/etc.
#------------------------------------------

def readwav(wav_path):
    wav, sr = sf.read(wav_path, dtype=np.float32)
    if len(wav.shape) == 2:
        wav = wav.mean(1)
    if sr != 16000:
        wlen = int(wav.shape[0] / sr * 16000)
        wav = signal.resample(wav, wlen)
    return wav


#convert frame-numbers to timestamps in seconds
# w2v2 step size is about 20ms, or 50 frames per second
def f2s(fr):
	return fr/50

def fmt(frame_aligns):
	return [(label,f2s(start),f2s(end)) for label,start,end in frame_aligns]

# generate mfa format for character (phone) and word alignments
def mfalike(chars,wds):
	hed = ['Begin,End,Label,Type,Speaker\n']
	wlines = [f'{s},{e},{w},words,000\n' for w,s,e in wds]
	slines = [f'{s},{e},{sg},phones,000\n' for sg,s,e in chars]
	return (''.join(hed+wlines+slines))


# prepare the input transcript text string
# TODO: 
# handle input strings that still have punctuation,
# or that have characters not present in labels_dict
def prep_transcript(xcp,lang):
    xcp = xcp.lower()
    while '  ' in xcp:
        xcp = xcp.replace('  ', ' ')
    xcp = xcp.replace(' ',d[lang]['model_word_separator'])
    label_ids = [d[lang]['labels_dict'][c] for c in xcp]
    return xcp, label_ids


def langsalign(wav_path,transcript_string,lang):
	norm_txt, rec_label_ids = prep_transcript(transcript_string, lang)
	emit = get_frame_probs(wav_path, lang)
	trellis = get_trellis(emit, rec_label_ids, d[lang]['blank_id'])
	path = backtrack(trellis, emit, rec_label_ids, d[lang]['blank_id'])
	segments = merge_repeats(path,norm_txt)
	words = merge_words(segments, d[lang]['model_word_separator'])
	
	#segments = [s for s in segments if s[0] != model_word_separator]
	return mfalike(fmt(segments), fmt(words))