open_ai_rust 0.2.14

Open AI SDK for Rust. To my knowledge, the only fully comprehensive and up-to-date Open AI crate built in and for Rust. Provides both low-level control with high level ergonomics for doing cool things (the whole reason we use Rust in the first place). Is maintained and has been used and tested in products used in production.
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43

use reqwest::multipart;
use serde::Deserialize;
use crate::helpers::{get_key, get_url};


#[derive(Deserialize)]
struct WhisperResponse {
    text: String,
}

pub async fn open_ai_speech_to_text(
    mp3_bytes: Vec<u8>,
) -> Result<String, String> {
    let client = reqwest::Client::new();

    let part = multipart::Part::bytes(mp3_bytes)
        .file_name("audio.mp3")
        .mime_str("audio/mpeg")
        .map_err(|e| format!("Error creating multipart part: {}", e))?;

    let form = multipart::Form::new()
        .text("model", "gpt-4o-mini-transcribe")
        .part("file", part);

    let url = get_url("/v1/audio/transcriptions")?;

    let res = client
        .post(url)
        .bearer_auth(get_key()?)
        .multipart(form)
        .send()
        .await
        .map_err(|e| format!("Error sending request to OpenAI: {}", e))?;

    if !res.status().is_success() {
        let err_text = res.text().await.map_err(|e| format!("Error reading error response: {}", e))?;
        return Err(format!("OpenAI API error: {}", err_text));
    }

    let transcript: WhisperResponse = res.json().await.map_err(|e| format!("Error parsing response: {}", e))?;

    Ok(transcript.text)
}