Skip to content

examples/server: implement "verbose_json" format with token details - #1781

Merged
ggerganov merged 3 commits into
ggml-org:masterfrom
rmmh:server-verbose
Jan 18, 2024
Merged

examples/server: implement "verbose_json" format with token details#1781
ggerganov merged 3 commits into
ggml-org:masterfrom
rmmh:server-verbose

Conversation

@rmmh

@rmmh rmmh commented Jan 18, 2024

Copy link
Copy Markdown
Contributor

This is intended to mirror the format of openai's Python whisper.transcribe() output.

Sample usage and output:

$ curl 127.0.0.1:8080/inference -H "Content-Type: multipart/form-data" -F file="@smallstep.wav" -F response_format="verbose_json" | jq .
{
  "text": " That's one small step for man.\n One giant leap for mankind.\n",
  "segments": [
    {
      "id": 0,
      "text": " That's one small step for man.",
      "start": 0,
      "end": 5,
      "tokens": [
        1320,
        338,
        530,
        1402,
        2239,
        329,
        582,
        13
      ],
      "words": [
        {
          "word": " That",
          "start": 0.16,
          "end": 0.31,
          "probability": 0.31418749690055847
        },
        {
          "word": "'s",
          "start": 0.38,
          "end": 0.49,
          "probability": 0.8378852009773254
        },
...

@rmmh rmmh changed the title implement "verbose_json" format with token details examples/server: implement "verbose_json" format with token details Jan 18, 2024
@ggerganov
ggerganov merged commit c0329ac into ggml-org:master Jan 18, 2024
@ggerganov ggerganov mentioned this pull request Jan 27, 2024
Merged
@JacobLinCool JacobLinCool mentioned this pull request Jan 30, 2024
Merged
@ogtayhuseynov0

ogtayhuseynov0 commented Feb 5, 2024

Copy link
Copy Markdown

I was planning to implement it, thank you for this contribution.

jiahansu pushed a commit to WiseSync/whisper.cpp that referenced this pull request Apr 17, 2024
@rmmh @jiahansu
server : implement "verbose_json" format with token details (ggml-org…
7b9e641 
…#1781)

* examples/server: implement "verbose_json" format with token details.

This is intended to mirror the format of openai's Python
whisper.transcribe() return values.

* server: don't write WAV to a temporary file if not converting

* server: use std::lock_guard instead of manual lock/unlock
viktor-silakov pushed a commit to viktor-silakov/whisper_node_mic.cpp that referenced this pull request May 11, 2024
@rmmh
server : implement "verbose_json" format with token details (ggml-org…
42afd55 
…#1781)

* examples/server: implement "verbose_json" format with token details.

This is intended to mirror the format of openai's Python
whisper.transcribe() return values.

* server: don't write WAV to a temporary file if not converting

* server: use std::lock_guard instead of manual lock/unlock
iThalay pushed a commit to iThalay/whisper.cpp that referenced this pull request Sep 23, 2024
@rmmh @iThalay
server : implement "verbose_json" format with token details (ggml-org…
c687d03 
…#1781)

* examples/server: implement "verbose_json" format with token details.

This is intended to mirror the format of openai's Python
whisper.transcribe() return values.

* server: don't write WAV to a temporary file if not converting

* server: use std::lock_guard instead of manual lock/unlock
iThalay pushed a commit to iThalay/whisper.cpp that referenced this pull request Sep 23, 2024
@rmmh @iThalay
server : implement "verbose_json" format with token details (ggml-org…
a0f33ae 
…#1781)

* examples/server: implement "verbose_json" format with token details.

This is intended to mirror the format of openai's Python
whisper.transcribe() return values.

* server: don't write WAV to a temporary file if not converting

* server: use std::lock_guard instead of manual lock/unlock
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@ggerganov ggerganov ggerganov approved these changes

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

3 participants

@rmmh @ogtayhuseynov0 @ggerganov