Skip to content

Add fields to verbose_json response and show examples on the home page #1802

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
merged 4 commits into from
Jan 30, 2024
Merged

Add fields to verbose_json response and show examples on the home page #1802

Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
7689841
server: include additional fields in the verbose_json response as Ope…
JacobLinCool Jan 23, 2024
ba13056
server: show request examples on home page
JacobLinCool Jan 23, 2024
7161762
server: todo note for compression_ratio and no_speech_prob
JacobLinCool Jan 29, 2024
272506f
server: add simple demo form to the homepage
JacobLinCool Jan 29, 2024
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
89 changes: 87 additions & 2 deletions examples/server/server.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -543,7 +543,76 @@ int main(int argc, char ** argv) {
{"Access-Control-Allow-Origin", "*"},
{"Access-Control-Allow-Headers", "content-type"}});

std::string const default_content = "<html>hello</html>";
std::string const default_content = R"(
<html>
<head>
<title>Whisper.cpp Server</title>
<meta charset="utf-8">
<meta name="viewport" content="width=device-width">
<style>
body {
font-family: sans-serif;
}
form {
display: flex;
flex-direction: column;
align-items: flex-start;
}
label {
margin-bottom: 0.5rem;
}
input, select {
margin-bottom: 1rem;
}
button {
margin-top: 1rem;
}
</style>
</head>
<body>
<h1>Whisper.cpp Server</h1>

<h2>/inference</h2>
<pre>
curl 127.0.0.1:)" + std::to_string(sparams.port) + R"(/inference \
-H "Content-Type: multipart/form-data" \
-F file="@&lt;file-path&gt;" \
-F temperature="0.0" \
-F temperature_inc="0.2" \
-F response_format="json"
</pre>

<h2>/load</h2>
<pre>
curl 127.0.0.1:)" + std::to_string(sparams.port) + R"(/load \
-H "Content-Type: multipart/form-data" \
-F model="&lt;path-to-model-file&gt;"
</pre>

<div>
<h2>Try it out</h2>
<form action="/inference" method="POST" enctype="multipart/form-data">
<label for="file">Choose an audio file:</label>
<input type="file" id="file" name="file" accept="audio/*" required><br>

<label for="temperature">Temperature:</label>
<input type="number" id="temperature" name="temperature" value="0.0" step="0.01" placeholder="e.g., 0.0"><br>

<label for="response_format">Response Format:</label>
<select id="response_format" name="response_format">
<option value="verbose_json">Verbose JSON</option>
<option value="json">JSON</option>
<option value="text">Text</option>
<option value="srt">SRT</option>
<option value="vtt">VTT</option>
</select><br>

<button type="submit">Submit</button>
</form>
</div>
</body>
</html>
)";

// store default params so we can reset after each inference request
whisper_params default_params = params;
Expand Down Expand Up @@ -787,7 +856,13 @@ int main(int argc, char ** argv) {
} else if (params.response_format == vjson_format) {
/* try to match openai/whisper's Python format */
std::string results = output_str(ctx, params, pcmf32s);
json jres = json{{"text", results}};
json jres = json{
{"task", params.translate ? "translate" : "transcribe"},
{"language", whisper_lang_str_full(whisper_full_lang_id(ctx))},
{"duration", float(pcmf32.size())/WHISPER_SAMPLE_RATE},
{"text", results},
{"segments", json::array()}
};
const int n_segments = whisper_full_n_segments(ctx);
for (int i = 0; i < n_segments; ++i)
{
Expand All @@ -801,6 +876,7 @@ int main(int argc, char ** argv) {
segment["end"] = whisper_full_get_segment_t1(ctx, i) * 0.01;
}

float total_logprob = 0;
const int n_tokens = whisper_full_n_tokens(ctx, i);
for (int j = 0; j < n_tokens; ++j) {
whisper_token_data token = whisper_full_get_token_data(ctx, i, j);
Expand All @@ -815,8 +891,17 @@ int main(int argc, char ** argv) {
word["end"] = token.t1 * 0.01;
}
word["probability"] = token.p;
total_logprob += token.plog;
segment["words"].push_back(word);
}

segment["temperature"] = params.temperature;
segment["avg_logprob"] = total_logprob / n_tokens;

// TODO compression_ratio and no_speech_prob are not implemented yet
// segment["compression_ratio"] = 0;
// segment["no_speech_prob"] = 0;

jres["segments"].push_back(segment);
}
res.set_content(jres.dump(-1, ' ', false, json::error_handler_t::replace),
Expand Down