> ## Documentation Index
> Fetch the complete documentation index at: https://phaseo.app/docs/llms.txt
> Use this file to discover all available pages before exploring further.

> ## Agent Instructions
> Phaseo Gateway base URL: https://api.phaseo.app/v1. Send `Authorization: Bearer $PHASEO_API_KEY` with every request.
> Model IDs look like `provider/model`, for example `openai/gpt-6-astra`. Do not guess IDs: list them with `GET /v1/models` or check https://phaseo.app/models. `phaseo/free` routes to an available free model and needs no wallet balance.
> For new text integrations prefer `POST /v1/responses`. `POST /v1/chat/completions` (OpenAI format) and `POST /v1/messages` (Anthropic format) suit existing code.
> The full API contract is https://phaseo.app/docs/openapi/v1/openapi.yaml. Official SDKs: `@phaseo/sdk` (TypeScript) and `phaseo` (Python), plus Go, C#, PHP, Ruby and Rust.
> Search these docs over MCP at https://phaseo.app/docs/mcp, and install Phaseo skills with `npx skills add https://phaseo.app/docs`. If a page is wrong or out of date, report it with the MCP `submit_feedback` tool.
> Manage Phaseo accounts with the `phaseo` CLI (`npm install -g @phaseo/cli`, then `phaseo login` or `phaseo login --device-code`), the `/v1` Management API or the MCP server at `https://mcp.phaseo.app/mcp?access=manage`. Do not send users to the dashboard for account tasks. A person is still needed for sign-up, approving a sign-in, paying the first top-up on the checkout link that `phaseo credits add` returns when no card is saved, confirming each saved-card top-up from the emailed link, and account security (password, MFA, passkeys). Create API keys with `phaseo keys create --name <name> --json`; never guess one. Credit top-ups need `phaseo login --scopes billing:write` first. Task map: https://phaseo.app/docs/v1/developers/manage-from-an-agent.md.

# STT（语音转文本）

> 在任意 Phaseo SDK 中使用 POST /v1/audio/transcriptions 转写音频。

以带有 `file` 字段的 multipart 表单把音频上传到 `POST /v1/audio/transcriptions`。带 base64 字符串的 JSON 正文无法通过验证。只有 TypeScript SDK 能发送这种表单；在其他语言中，请像下面那样用标准 HTTP 客户端提交，或者改为发送 URL（参见[从 URL 转写音频](#transcribe-audio-from-a-url)）。

## 上传音频文件

```typescript TypeScript theme={null}
import { readFile } from "node:fs/promises";

const form = new FormData();
form.append("model", "openai/gpt-4o-transcribe");
form.append("language", "en");
form.append("file", new Blob([await readFile("call.mp3")], { type: "audio/mpeg" }), "call.mp3");

const transcript = await client.generateTranscription(form as unknown as Record<string, unknown>);
console.log(transcript.text);
```

```python Python theme={null}
# generate_transcription() 目前还不能发送 multipart，所以用 HTTPX 提交表单。
import os
import httpx

with open("call.mp3", "rb") as audio:
    response = httpx.post(
        "https://api.phaseo.app/v1/audio/transcriptions",
        headers={"Authorization": f"Bearer {os.environ['PHASEO_API_KEY']}"},
        data={"model": "openai/gpt-4o-transcribe", "language": "en"},
        files={"file": ("call.mp3", audio, "audio/mpeg")},
        timeout=120,
    )
response.raise_for_status()
print(response.json()["text"])
```

```go Go theme={null}
// CreateTranscription() 发送的是 JSON，所以用 net/http 和 mime/multipart 提交表单。
var body bytes.Buffer
form := multipart.NewWriter(&body)
_ = form.WriteField("model", "openai/gpt-4o-transcribe")
_ = form.WriteField("language", "en")
part, _ := form.CreateFormFile("file", "call.mp3")
audio, _ := os.ReadFile("call.mp3")
_, _ = part.Write(audio)
_ = form.Close()

req, _ := http.NewRequest("POST", "https://api.phaseo.app/v1/audio/transcriptions", &body)
req.Header.Set("Authorization", "Bearer "+os.Getenv("PHASEO_API_KEY"))
req.Header.Set("Content-Type", form.FormDataContentType())

resp, err := http.DefaultClient.Do(req)
```

```csharp C# theme={null}
// CreateTranscription() 发送的是 JSON，所以用 HttpClient 提交表单。
using var http = new HttpClient();
http.DefaultRequestHeaders.Authorization =
    new("Bearer", Environment.GetEnvironmentVariable("PHASEO_API_KEY"));

using var form = new MultipartFormDataContent
{
    { new StringContent("openai/gpt-4o-transcribe"), "model" },
    { new StringContent("en"), "language" },
    { new ByteArrayContent(await File.ReadAllBytesAsync("call.mp3")), "file", "call.mp3" },
};

var response = await http.PostAsync("https://api.phaseo.app/v1/audio/transcriptions", form);
Console.WriteLine(await response.Content.ReadAsStringAsync());
```

```php PHP theme={null}
<?php
// generateTranscription() 发送的是 JSON，所以用 Guzzle 提交表单；Composer 安装 SDK 时已一并安装 Guzzle。
require 'vendor/autoload.php';

$http = new GuzzleHttp\Client();

$response = $http->post('https://api.phaseo.app/v1/audio/transcriptions', [
    'headers' => ['Authorization' => 'Bearer ' . getenv('PHASEO_API_KEY')],
    'multipart' => [
        ['name' => 'model', 'contents' => 'openai/gpt-4o-transcribe'],
        ['name' => 'language', 'contents' => 'en'],
        ['name' => 'file', 'contents' => fopen('call.mp3', 'r'), 'filename' => 'call.mp3'],
    ],
]);

echo $response->getBody();
```

```ruby Ruby theme={null}
# generate_transcription 发送的是 JSON，所以用 Net::HTTP 提交表单。
require 'net/http'

uri = URI('https://api.phaseo.app/v1/audio/transcriptions')
request = Net::HTTP::Post.new(uri)
request['Authorization'] = "Bearer #{ENV.fetch('PHASEO_API_KEY')}"
request.set_form(
  [['model', 'openai/gpt-4o-transcribe'], ['language', 'en'], ['file', File.open('call.mp3')]],
  'multipart/form-data'
)

response = Net::HTTP.start(uri.host, uri.port, use_ssl: true) { |http| http.request(request) }
puts response.body
```

```java Java theme={null}
// SDK 的 JSON 辅助方法不编码 multipart，所以用 java.net.http 提交表单。
import java.io.IOException;
import java.net.URI;
import java.net.http.HttpClient;
import java.net.http.HttpRequest;
import java.net.http.HttpRequest.BodyPublishers;
import java.net.http.HttpResponse.BodyHandlers;
import java.nio.file.Path;
import java.time.Duration;
import java.util.UUID;

String baseUrl = System.getenv().getOrDefault("PHASEO_BASE_URL", "https://api.phaseo.app/v1")
    .replaceAll("/+$", "");
String apiKey = System.getenv("PHASEO_API_KEY");
String boundary = "phaseo-" + UUID.randomUUID();
var body = BodyPublishers.concat(
    BodyPublishers.ofString("--" + boundary + "\r\nContent-Disposition: form-data; name=\"model\"\r\n\r\nopenai/gpt-4o-transcribe\r\n"),
    BodyPublishers.ofString("--" + boundary + "\r\nContent-Disposition: form-data; name=\"language\"\r\n\r\nen\r\n"),
    BodyPublishers.ofString("--" + boundary + "\r\nContent-Disposition: form-data; name=\"file\"; filename=\"call.mp3\"\r\nContent-Type: audio/mpeg\r\n\r\n"),
    BodyPublishers.ofFile(Path.of("call.mp3")),
    BodyPublishers.ofString("\r\n--" + boundary + "--\r\n")
);
var request = HttpRequest.newBuilder(URI.create(baseUrl + "/audio/transcriptions"))
    .header("Authorization", "Bearer " + apiKey)
    .header("Content-Type", "multipart/form-data; boundary=" + boundary)
    .timeout(Duration.ofSeconds(60))
    .POST(body)
    .build();
var response = HttpClient.newHttpClient().send(request, BodyHandlers.ofString());
if (response.statusCode() < 200 || response.statusCode() >= 300) {
    throw new IOException("HTTP " + response.statusCode() + ": " + response.body());
}
System.out.println(response.body());
```

| 语言 | SDK 方法 | 能上传文件吗？ |
| - | - | - |
| TypeScript | `client.generateTranscription(form)` | 能。传入 `FormData` 正文，SDK 会原样以 multipart 发送。需要类型转换，是因为该方法的类型描述的是 JSON 正文。 |
| Python | `client.generate_transcription(request)` | 不能。它总是发送 JSON，所以请像上面那样用 HTTPX 提交表单。 |
| Go | `client.CreateTranscription(ctx, req)` | 不能。它发送 JSON；请使用 `net/http` 和 `mime/multipart`。 |
| C# | `client.CreateTranscription(request)` | 不能。它发送 JSON；请使用 `HttpClient` 和 `MultipartFormDataContent`。 |
| PHP | `$client->generateTranscription($request)`，别名 `createTranscription()` | 不能。它发送 JSON；请使用 Guzzle 的 `multipart` 选项。 |
| Ruby | `client.generate_transcription(...)`，别名 `create_transcription` | 不能。它发送 JSON；请使用 `Net::HTTP` 和 `set_form`。 |
| Java | 无 | 不能。请使用 `java.net.http.HttpClient` 发送 multipart 正文。 |
| Rust | 无 | 不能。`client.post()` 只发送 JSON，因此只适用于 [URL 来源](#transcribe-audio-from-a-url)。 |

## 从 URL 转写音频

Mistral、ElevenLabs 和 Morpheus 模型可以从 URL 读取音频，而不需要上传。由于这种请求是 JSON，无法上传文件的 SDK 方法也能处理它：

```python Python theme={null}
result = client.generate_transcription({
    "model": "mistral/voxtral-mini-transcribe-2",
    "file_url": "https://example.com/call.mp3",
    "language": "en",
})

print(result["text"])
```

```go Go theme={null}
resp, err := client.CreateTranscription(context.Background(), map[string]interface{}{
  "model":    "mistral/voxtral-mini-transcribe-2",
  "file_url": "https://example.com/call.mp3",
  "language": "en",
})
if err != nil {
  panic(err)
}
fmt.Println(resp["text"])
```

```csharp C# theme={null}
var response = await client.CreateTranscription(new Dictionary<string, object>
{
    ["model"] = "mistral/voxtral-mini-transcribe-2",
    ["file_url"] = "https://example.com/call.mp3",
    ["language"] = "en",
});
```

```php PHP theme={null}
$response = $client->generateTranscription([
    'model' => 'mistral/voxtral-mini-transcribe-2',
    'file_url' => 'https://example.com/call.mp3',
    'language' => 'en',
]);

echo $response['text'];
```

```ruby Ruby theme={null}
response = client.generate_transcription(
  model: 'mistral/voxtral-mini-transcribe-2',
  file_url: 'https://example.com/call.mp3',
  language: 'en'
)
```

```rust Rust theme={null}
let response = client.post("/audio/transcriptions", &json!({
    "model": "mistral/voxtral-mini-transcribe-2",
    "file_url": "https://example.com/call.mp3",
    "language": "en"
}))?;

println!("{}", response.body["text"]);
```

## 参数

在每种语言中，字段名都是表单字段（使用 URL 来源时为 JSON 字段）。

| 字段 | 说明 |
| - | - |
| `model` | 必填。转写模型 ID。 |
| `file` | 上传的音频：`aac`、`flac`、`mp3`、`mp4`、`mpeg`、`mpga`、`m4a`、`ogg`、`opus`、`wav` 或 `webm`。除使用下面的 URL 选项外均为必填。大多数模型最大 25 MB。 |
| `file_url`、`file_id`、`s3_presigned_url` | Mistral、ElevenLabs 和 Morpheus 模型只接受一个音频来源，因此可以用 URL 代替 `file`。 |
| `language` | 语言提示，例如 `en`。 |
| `prompt` | 用于提高准确度的上下文。 |
| `temperature` | `0` 到 `1`（ElevenLabs 为 `0` 到 `2`）。 |
| `response_format` | `json`、`text`、`srt`、`verbose_json`、`vtt` 或 `diarized_json`。 |
| `timestamp_granularities` | `word` 和/或 `segment`。 |
| `stream` | 在模型支持时，边生成边流式输出转写文本。 |

完整的 schema 见 [Transcriptions API 参考](https://phaseo.app/docs/zh-Hans/v1/api-reference/endpoint/audio-transcriptions.md)。

## 响应

`AudioTranscriptionResponse`，以各 SDK 的映射类型返回（Go 中为 `map[string]interface{}`，C# 中为 `Dictionary<string, object>?`，PHP 中为数组，Ruby 中为 Hash）：

```json theme={null}
{
  "text": "Hello world"
}
```


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.