|
1 | 1 | # Whisper-API-unity |
2 | | -Binds [Whisper transcription and translation API](https://platform.openai.com/docs/api-reference/audio) to pure C# on Unity. |
| 2 | +A client library of OpenAI [Whisper transcription and translation API](https://platform.openai.com/docs/api-reference/audio) for Unity. |
3 | 3 |
|
4 | 4 | See also [official document](https://platform.openai.com/docs/guides/speech-to-text). |
5 | 5 |
|
6 | | -## How to import by UnityPackageManager |
| 6 | +## Features |
7 | 7 |
|
8 | | -Add dependencies: |
| 8 | +- Transcription |
| 9 | + - Speech audio file to text in speeched language. |
| 10 | +- Translation |
| 11 | + - Speech audio file to text in English. |
| 12 | + |
| 13 | +## How to import by Unity Package Manager |
| 14 | + |
| 15 | +Add following dependencies to your `/Packages/manifest.json`. |
9 | 16 |
|
10 | 17 | ```json |
11 | 18 | { |
12 | 19 | "dependencies": { |
13 | | - "com.mochineko.whisper-api": "https://github.com/mochi-neko/Whisper-API-unity.git?path=/Assets/Mochineko/Whisper_API#0.1.0", |
14 | | - "com.unity.nuget.newtonsoft-json": "3.0.2", |
| 20 | + "com.mochineko.whisper-api": "https://github.com/mochi-neko/Whisper-API-unity.git?path=/Assets/Mochineko/Whisper_API#1.0.0", |
15 | 21 | ... |
16 | 22 | } |
17 | 23 | } |
18 | 24 | ``` |
19 | 25 |
|
20 | | -to your `mainfest.json`. |
21 | | - |
22 | | -If you have already used Newtonsoft.Json on your project, remove dependency:`"com.unity.nuget.newtonsoft-json": "3.0.2",`. |
23 | | - |
24 | | -## How to use transcription speech audio into text by Whisper API |
25 | | - |
26 | | -1. Generate API key on [OpenAI](https://platform.openai.com/account/api-keys). (Take care your API key, this is a secret information then you should not open.) |
27 | | -2. Create an instance of `WhisperTranscriptionConnection` with API key. |
28 | | -3. Set file path of speech audio e.g. `/some/path/of/file/audio.mp3`, and call `WhisperTranscriptionConnection.TranscribeFromFileAsync`. |
29 | | -4. Because deault response text format is JSON, you can exclude text by `APIResponseBody.FromJson(json)?.Text`. |
| 26 | +## How to use |
30 | 27 |
|
31 | | -An essential sample code of transcription with [UniTask](https://github.com/Cysharp/UniTask) is as follows: |
32 | | - |
33 | | -```cs |
34 | | -#nullable enable |
35 | | -using System; |
36 | | -using Cysharp.Threading.Tasks; |
37 | | -using UnityEngine; |
38 | | -using UnityEngine.Assertions; |
39 | | -using Mochineko.Whisper_API.; |
40 | | -using Mochineko.Whisper_API.Transcription; |
41 | | - |
42 | | -namespace XXX |
43 | | -{ |
44 | | - /// <summary> |
45 | | - /// A sample component to transcribe speech by Whisper API on Unity. |
46 | | - /// </summary> |
47 | | - public sealed class TranscriptionSample : MonoBehaviour |
48 | | - { |
49 | | - /// <summary> |
50 | | - /// API key generated by OpenAPI. |
51 | | - /// </summary> |
52 | | - [SerializeField] private string apiKey = string.Empty; |
53 | | - |
54 | | - /// <summary> |
55 | | - /// File path of speech audio. |
56 | | - /// </summary> |
57 | | - [SerializeField] private string filePath = string.Empty; |
58 | | - |
59 | | - private WhisperTranscriptionConnection? connection; |
60 | | - |
61 | | - private void Start() |
62 | | - { |
63 | | - // Create instance of WhisperTranscriptionConnection. |
64 | | - connection = new WhisperTranscriptionConnection(apiKey); |
65 | | - |
66 | | - // If you want to specify response format, language, etc..., please use other initialization: |
67 | | - // connection = new WhisperTranscriptionConnection(apiKey, new APIRequestBody( |
68 | | - // file: "", |
69 | | - // model: "whisper-1", |
70 | | - // prompt: "Some prompts", |
71 | | - // responseFormat: "json", |
72 | | - // temperature: 1f, |
73 | | - // language: "ja")); |
74 | | - } |
75 | | - |
76 | | - [ContextMenu(nameof(Transcribe))] |
77 | | - public async void Transcribe() |
78 | | - { |
79 | | - string result; |
80 | | - try |
81 | | - { |
82 | | - // Transcribe speech by Whisper speech to text API. |
83 | | - result = await connection |
84 | | - .TranscribeFromFileAsync(filePath, this.GetCancellationTokenOnDestroy()); |
85 | | - } |
86 | | - catch (Exception e) |
87 | | - { |
88 | | - // Exceptions should be caught. |
89 | | - Debug.LogException(e); |
90 | | - return; |
91 | | - } |
92 | | - |
93 | | - // Default text response format is JSON. |
94 | | - var text = APIResponseBody.FromJson(result)?.Text; |
95 | | - |
96 | | - // Log text result. |
97 | | - Debug.Log($"[Whisper_API.Transcription.Samples] Result:\n{text}"); |
98 | | - } |
99 | | - } |
100 | | -} |
101 | | -``` |
| 28 | +Please generate your API key on [OpenAI](https://platform.openai.com/account/api-keys). |
102 | 29 |
|
103 | | -See also [Sample](https://github.com/mochi-neko/Whisper-API-unity/blob/main/Assets/Mochineko/Whisper_API.Samples/Transcription/TranscriptionSample.cs). |
| 30 | +See [sample codes](./Assets/Mochineko/WhisperAPI.Samples). |
104 | 31 |
|
105 | | -Translation (translate speech audio into English text) sample is [here](https://github.com/mochi-neko/Whisper-API-unity/blob/main/Assets/Mochineko/Whisper_API.Samples/Transcription/TranscriptionSample.cs). |
| 32 | +You can customize handling of retries to fit needs of your project by [Relent](https://github.com/mochi-neko/Relent), |
| 33 | +e.g. [PolicyFactory](./Assets/Mochineko/WhisperAPI.Samples/PolicyFactory.cs). |
106 | 34 |
|
107 | 35 | ## Changelog |
108 | 36 |
|
109 | | -See [CHANGELOG](https://github.com/mochi-neko/Whisper-API-unity/blob/main/CHANGELOG.md) |
| 37 | +See [CHANGELOG](./CHANGELOG.md). |
110 | 38 |
|
111 | | -## 3rd Party Notices |
| 39 | +## 3rd party notices |
112 | 40 |
|
113 | | -See [NOTICE](https://github.com/mochi-neko/Whisper-API-unity/blob/main/NOTICE.md). |
| 41 | +See [NOTICE](./NOTICE.md). |
114 | 42 |
|
115 | 43 | ## License |
116 | 44 |
|
117 | | -[MIT License](https://github.com/mochi-neko/Whisper-API-unity/blob/main/LICENSE) |
| 45 | +Licensed under the [MIT](./LICENSE) License. |
0 commit comments