ggml-org/whisper.cpp

Streaming to file leads to "duplicates"

Open

#1,056 opened on Jun 27, 2023

View on GitHub
 (3 comments) (0 reactions) (0 assignees)C++ (5,535 forks)batch import
enhancementgood first issue

Repository metrics

Stars
 (49,693 stars)
PR merge metrics
 (Avg merge 5d 20h) (36 merged PRs in 30d)

Description

Hi all,

First of all, amazing work on this project. I'm really excited about what you're doing.

There's something I noticed about the way the ./stream works with the file parameter. I believe that for the stdout, it's only printing to a newline every n_new_line, and it would otherwise use a carriage return to overwrite the last line. This works nicely for stdout, but with the file output parameter, it's printed as a new line for every iteration.

I think this is buggy because the resulting text file has a bunch of duplicate lines most of the time, and it's difficult to parse out later. I'm guessing that this was written this way because you can't write carriage returns to a file and have it clear the last line like you can with stdout, but I think it would be a good compromise to only write to file on every n_new_line.

What do you think?

Contributor guide

Research direction
Examine the streaming output logic in the codebase, specifically how file vs stdout output is handled. Look for the use of carriage returns and newlines in the output loop, and consider modifying the file output to only write at every n new line.
Tech stack
ccpp
Domain
backendcli
Issue type
Bug
DifficultyEstimated implementation difficulty for a new contributor, from 1 for very small changes to 5 for expert-level work.
2
Estimated timeA rough time range for an experienced contributor to investigate, implement, test, and prepare a pull request.
1-3 hours
Activity statusHow available the issue appears right now: fresh, active, stale, blocked, or waiting on maintainer input.
Active
ClarityHow clearly the issue explains the expected change, acceptance criteria, and next step.
Mostly clear
Prerequisites
GitC++
Newbie friendlinessA 1-100 score estimating how approachable this issue is for first-time contributors.
75