ctcalign.py

from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor
import torch
import numpy as np
from dataclasses import dataclass




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

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

# important to know for CTC decoding - potentially language/model dependent
#model_word_separator = '|'
#model_blank_token = '[PAD]'
#is_MODEL_PATH="../models/LVL/wav2vec2-large-xlsr-53-icelandic-ep10-1000h"



class CTCAligner:
    
    def __init__(self, model_path,model_word_separator, model_blank_token):
        self.device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
        torch.random.manual_seed(0)
        
        self.model = Wav2Vec2ForCTC.from_pretrained(model_path).to(self.device)
        self.processor = Wav2Vec2Processor.from_pretrained(model_path)

        # build labels dict from a processor where it is not directly accessible
        max_labels = 100 # any reasonable number higher than vocab + extra + special tokens in any language used
        ixs = sorted(list(range(max_labels)),reverse=True)
        self.labels_dict = {self.processor.tokenizer.decode(n) or model_word_separator:n for n in ixs}
        
        self.blank_id = self.labels_dict[model_blank_token]
        self.model_word_separator = model_word_separator





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


# return the label class probability of each audio frame
# wav is the wav data already read in, NOT the file path.
def get_frame_probs(wav,aligner):
    with torch.inference_mode(): # similar to with torch.no_grad():
        input_values = aligner.processor(wav,sampling_rate=16000).input_values[0]
        input_values = torch.tensor(input_values, device=aligner.device).unsqueeze(0)
        emits =  aligner.model(input_values).logits
        emits = torch.log_softmax(emits, dim=-1)
    return emits[0].cpu().detach()


def get_trellis(emission, tokens, blank_id):
    
    num_frame = emission.size(0)
    num_tokens = len(tokens)
    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.maximum(
            # 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



@dataclass
class Point:
    token_index: int
    time_index: int
    score: float
    
@dataclass
class Segment:
    label: str
    start: int
    end: int
    score: float

    @property
    def mfaform(self):
        return f"{f2s(self.start)},{f2s(self.end)},{self.label}"

    @property
    def length(self):
        return self.end - self.start
    
    
    
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
        # `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(Point(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].token_index == path[i2].token_index: # while both path steps point to the same token index
            i2 += 1
        score = sum(path[k].score for k in range(i1, i2)) / (i2 - i1)
        segments.append( # when i2 finally switches to a different token,
            Segment(
                transcript[path[i1].token_index],# to the list of segments, append the token from i1
                path[i1].time_index, # time of the first path-point of that token
                path[i2 - 1].time_index + 1, # time of the final path-point for that token.
                score,
            )
        )
        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].label == separator:
            if i1 != i2:
                segs = segments[i1:i2]
                word = "".join([seg.label for seg in segs])
                score = sum(seg.score * seg.length for seg in segs) / sum(seg.length for seg in segs)
                words.append(Segment(word, segments[i1].start, segments[i2 - 1].end, score))
            i1 = i2 + 1
            i2 = i1
        else:
            i2 += 1
    return words



#------------------------------------------
# handle etc.
#------------------------------------------


# generate mfa format for character (phone) and word alignments
# skip the word separator as it is not a phone
def mfalike(chars,wds,wsep):
	hed = ['Begin,End,Label,Type,Speaker\n']
	wlines = [f'{w.mfaform},words,000\n' for w in wds]
	slines = [f'{ch.mfaform},phones,000\n' for ch in chars if ch.label != wsep]
	return (''.join(hed+wlines+slines))
	
# generate basic exportable list format for character OR word alignments
# skip the word separator as it is not a phone
def basic(segs,wsep="|"):
    return [[s.label,f2s(s.start),f2s(s.end)] for s in segs if s.label != wsep]
	

# needs pad labels added to correctly time first segment
# and therefore add word sep character as placeholder in transcript
def prep_transcript(xcp, aligner):
    xcp = xcp.replace(' ', aligner.model_word_separator)
    label_ids = [aligner.labels_dict[c] for c in xcp]
    label_ids = [aligner.blank_id] +  label_ids + [aligner.blank_id]
    xcp = f'{ aligner.model_word_separator}{xcp}{aligner.model_word_separator}'
    return xcp,label_ids



def align(wav_data,transcript,aligner):
	norm_transcript,rec_label_ids = prep_transcript(transcript, aligner)
	emit = get_frame_probs(wav_data,aligner)
	trellis = get_trellis(emit, rec_label_ids,  aligner.blank_id)
	path = backtrack(trellis, emit, rec_label_ids,  aligner.blank_id)
	
	segments = merge_repeats(path,norm_transcript)
	words = merge_words(segments, aligner.model_word_separator)
	
	#segments = [s for s in segments if s[0] != model_word_separator]
	#return mfalike(segments,words,model_word_separator)
	return basic(words,aligner.model_word_separator), basic(segments,aligner.model_word_separator)