Skip to content

Fix DTW boundary guards and seek loop in transcribe_chunk #3

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
TroyHernandez merged 1 commit into from
Feb 27, 2026
Merged

Fix DTW boundary guards and seek loop in transcribe_chunk #3

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
20 changes: 13 additions & 7 deletions R/alignment.R
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -231,19 +231,25 @@ dtw_align <- function(cost) {
D[1, 1] <- cost[1, 1]

# First row: can only come from the left
for (j in 2:m) {
D[1, j] <- D[1, j - 1] + cost[1, j]
if (m >= 2L) {
for (j in 2:m) {
D[1, j] <- D[1, j - 1] + cost[1, j]
}
}

# First column: can only come from above
for (i in 2:n) {
D[i, 1] <- D[i - 1, 1] + cost[i, 1]
if (n >= 2L) {
for (i in 2:n) {
D[i, 1] <- D[i - 1, 1] + cost[i, 1]
}
}

# Fill rest
for (i in 2:n) {
for (j in 2:m) {
D[i, j] <- cost[i, j] + min(D[i - 1, j], D[i, j - 1], D[i - 1, j - 1])
if (n >= 2L && m >= 2L) {
for (i in 2:n) {
for (j in 2:m) {
D[i, j] <- cost[i, j] + min(D[i - 1, j], D[i, j - 1], D[i - 1, j - 1])
}
}
}

Expand Down
3 changes: 3 additions & 0 deletions R/transcribe.R
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -299,6 +299,8 @@ transcribe_chunk <- function(

if (seek == 0L) {
mel <- full_mel
} else if (seek + 1L > n_frames) {
break
} else {
# Slice mel[1, :, seek:] and pad to n_frames width
mel_slice <- full_mel[, , (seek + 1L):n_frames]
Expand Down Expand Up @@ -335,6 +337,7 @@ transcribe_chunk <- function(
device = device)

generated <- decode_result$tokens
all_generated <- c(all_generated, generated)

# Find the last timestamp token to determine where to seek next
last_ts_frame <- 0L
Expand Down