openai/openai-dotnet
Publicmirrored from https://github.com/openai/openai-dotnetAvailable
tests/Audio/TranslationTests.cs
120lines · modecode
| 1 | using NUnit.Framework; |
| 2 | using OpenAI.Audio; |
| 3 | using OpenAI.Tests.Utility; |
| 4 | using System; |
| 5 | using System.ClientModel; |
| 6 | using System.IO; |
| 7 | using System.Threading; |
| 8 | using System.Threading.Tasks; |
| 9 | using static OpenAI.Tests.TestHelpers; |
| 10 | |
| 11 | namespace OpenAI.Tests.Audio; |
| 12 | |
| 13 | [TestFixture(true)] |
| 14 | [TestFixture(false)] |
| 15 | [Parallelizable(ParallelScope.All)] |
| 16 | [Category("Audio")] |
| 17 | public partial class TranslationTests : SyncAsyncTestBase |
| 18 | { |
| 19 | public TranslationTests(bool isAsync) : base(isAsync) |
| 20 | { |
| 21 | } |
| 22 | |
| 23 | public enum AudioSourceKind |
| 24 | { |
| 25 | UsingStream, |
| 26 | UsingFilePath, |
| 27 | } |
| 28 | |
| 29 | [Test] |
| 30 | [TestCase(AudioSourceKind.UsingStream)] |
| 31 | [TestCase(AudioSourceKind.UsingFilePath)] |
| 32 | public async Task TranslationWorks(AudioSourceKind audioSourceKind) |
| 33 | { |
| 34 | AudioClient client = GetTestClient<AudioClient>(TestScenario.Audio_Whisper); |
| 35 | |
| 36 | string filename = "audio_french.wav"; |
| 37 | string path = Path.Combine("Assets", filename); |
| 38 | AudioTranslation translation = null; |
| 39 | |
| 40 | if (audioSourceKind == AudioSourceKind.UsingStream) |
| 41 | { |
| 42 | using FileStream audio = File.OpenRead(path); |
| 43 | |
| 44 | translation = IsAsync |
| 45 | ? await client.TranslateAudioAsync(audio, filename) |
| 46 | : client.TranslateAudio(audio, filename); |
| 47 | } |
| 48 | else if (audioSourceKind == AudioSourceKind.UsingFilePath) |
| 49 | { |
| 50 | translation = IsAsync |
| 51 | ? await client.TranslateAudioAsync(path) |
| 52 | : client.TranslateAudio(path); |
| 53 | } |
| 54 | Assert.That(translation?.Text, Is.Not.Null); |
| 55 | Assert.That(translation.Text.ToLowerInvariant(), Contains.Substring("whisper")); |
| 56 | } |
| 57 | |
| 58 | [Test] |
| 59 | [TestCase("text")] |
| 60 | [TestCase("json")] |
| 61 | [TestCase("verbose_json")] |
| 62 | [TestCase("srt")] |
| 63 | [TestCase("vtt")] |
| 64 | [TestCase(null)] |
| 65 | public async Task TranslationFormatsWork(string responseFormat) |
| 66 | { |
| 67 | AudioClient client = GetTestClient<AudioClient>(TestScenario.Audio_Whisper); |
| 68 | string path = Path.Combine("Assets", "audio_french.wav"); |
| 69 | |
| 70 | AudioTranslationOptions options = new() |
| 71 | { |
| 72 | ResponseFormat = responseFormat switch |
| 73 | { |
| 74 | "text" => AudioTranslationFormat.Text, |
| 75 | "json" => AudioTranslationFormat.Simple, |
| 76 | "verbose_json" => AudioTranslationFormat.Verbose, |
| 77 | "srt" => AudioTranslationFormat.Srt, |
| 78 | "vtt" => AudioTranslationFormat.Vtt, |
| 79 | _ => (AudioTranslationFormat?)null |
| 80 | } |
| 81 | }; |
| 82 | |
| 83 | AudioTranslation translation = IsAsync |
| 84 | ? await client.TranslateAudioAsync(path, options) |
| 85 | : client.TranslateAudio(path, options); |
| 86 | |
| 87 | Assert.That(translation?.Text?.ToLowerInvariant(), Does.Contain("recognition")); |
| 88 | |
| 89 | if (options.ResponseFormat == AudioTranslationFormat.Verbose) |
| 90 | { |
| 91 | Assert.That(translation.Language, Is.EqualTo("english")); |
| 92 | Assert.That(translation.Duration, Is.GreaterThan(TimeSpan.Zero)); |
| 93 | Assert.That(translation.Segments, Is.Not.Empty); |
| 94 | |
| 95 | for (int i = 0; i < translation.Segments.Count; i++) |
| 96 | { |
| 97 | TranscribedSegment segment = translation.Segments[i]; |
| 98 | |
| 99 | if (i > 0) |
| 100 | { |
| 101 | Assert.That(segment.StartTime, Is.GreaterThanOrEqualTo(translation.Segments[i - 1].EndTime)); |
| 102 | } |
| 103 | |
| 104 | Assert.That(segment.Id, Is.EqualTo(i)); |
| 105 | Assert.That(segment.EndTime, Is.GreaterThanOrEqualTo(segment.StartTime)); |
| 106 | Assert.That(segment.TokenIds.Span.Length, Is.GreaterThan(0)); |
| 107 | |
| 108 | Assert.That(segment.AverageLogProbability, Is.LessThan(-0.001f).Or.GreaterThan(0.001f)); |
| 109 | Assert.That(segment.CompressionRatio, Is.LessThan(-0.001f).Or.GreaterThan(0.001f)); |
| 110 | Assert.That(segment.NoSpeechProbability, Is.LessThan(-0.001f).Or.GreaterThan(0.001f)); |
| 111 | } |
| 112 | } |
| 113 | else |
| 114 | { |
| 115 | Assert.That(translation.Duration, Is.Null); |
| 116 | Assert.That(translation.Language, Is.Null); |
| 117 | Assert.That(translation.Segments, Is.Not.Null.And.Empty); |
| 118 | } |
| 119 | } |
| 120 | } |
| 121 | |