Name: RVC v2 API: AI Voice Conversion with Custom Models
Brand: magicapi
SKU: api-magicapi-ai-speech-to-speech-conversion-pseudoram-rvc-v2
Availability: InStock
Author: API Market

Shashank Agarwal

# 🎙️ Voice Conversion API (pseudoram RVC v2)

## Overview
RVC v2, published by pseudoram, converts speech or singing from one voice to another (speech to speech) with any voice model trained with RVC v2 (Retrieval-based Voice Conversion).

Send an `input_audio` URL and choose a voice: one of the built-in `rvc_model` voices, or your own RVC v2 model loaded from `custom_rvc_model_download_url`. The API returns the URL of the converted audio file. You can shift the pitch, choose the pitch detection method and control how much of the original voice's accent, loudness and breath is kept.

## Features
- **Speech to speech**: the words and timing of the input stay; the voice changes.
- **Built-in voices or your own model**: `rvc_model` accepts `Obama`, `Trump`, `Sandy`, `Rogan` or `CUSTOM`. A `custom_rvc_model_download_url` is downloaded and used whatever `rvc_model` says.
- **Pitch shift**: `pitch_change` in semitones (positive is higher, negative is lower).
- **Pitch detection**: `f0_method` `rmvpe` (default, for clarity in vocals) or `mangio-crepe` (smoother vocals), with `crepe_hop_length` for `mangio-crepe`.
- **Voice blending controls**: `index_rate` (accent), `rms_mix_rate` (loudness), `protect` (breath and voiceless consonants) and `filter_radius` (median filtering of pitch).
- **Output formats**: `mp3` (default, smaller files) or `wav` (best quality, larger files).

## Use Cases
1. **Voice-overs and dubbing**: re-voice a recording with a character voice you trained.
2. **Song covers**: convert a vocal track to another singer's voice model, with a pitch shift to match the range.
3. **Prototyping characters**: try voices for games and animation before recording.

## Visual Examples
### Example Input
Audio: [Listen to the example input](https://blog.api.market/wp-content/uploads/2026/10/rvc-v2-input-api-market.mp3)

Voice: a custom RVC v2 model loaded from `https://huggingface.co/Argax/doofenshmirtz-RUS/resolve/main/doofenshmirtz.zip` with `rvc_model` set to `CUSTOM`.

Settings: `protect` 0.5, `f0_method` `rmvpe`, `index_rate` 1, `pitch_change` 8, `rms_mix_rate` 1, `filter_radius` 1, `output_format` `mp3`, `crepe_hop_length` 128.

### Example Output
[Listen to the example output](https://blog.api.market/wp-content/uploads/2026/10/rvc-v2-output-2v12vrqn-api-market-1.wav)

The example asked for `mp3`, and the file the model returned was a WAV file.

## Input Parameters
| Field | Type | Required | Default | Allowed values | Description |
| --- | --- | --- | --- | --- | --- |
| `input_audio` | string (URL) | No | none |  | Audio file to convert |
| `rvc_model` | string | No | `Obama` | `Obama`, `Trump`, `Sandy`, `Rogan`, `CUSTOM` | RVC model for a specific voice. If using a custom model, this should match the name of the downloaded model. If a `custom_rvc_model_download_url` is provided, this is set automatically to the name of the downloaded model |
| `custom_rvc_model_download_url` | string | No | none |  | URL to download a custom RVC model. If provided, the model is downloaded (if it doesn't already exist) and used for prediction, regardless of the `rvc_model` value |
| `pitch_change` | number | No | `0` |  | Adjust pitch of AI vocals in semitones. Use positive values to increase pitch, negative to decrease |
| `index_rate` | number | No | `0.5` | 0 to 1 | Control how much of the AI's accent to leave in the vocals |
| `filter_radius` | integer | No | `3` | 0 to 7 | If >=3: apply median filtering to the harvested pitch results |
| `rms_mix_rate` | number | No | `0.25` | 0 to 1 | Control how much to use the original vocal's loudness (0) or a fixed loudness (1) |
| `f0_method` | string | No | `rmvpe` | `rmvpe`, `mangio-crepe` | Pitch detection algorithm. `rmvpe` for clarity in vocals, `mangio-crepe` for smoother vocals |
| `crepe_hop_length` | integer | No | `128` |  | When `f0_method` is set to `mangio-crepe`, this controls how often it checks for pitch changes in milliseconds |
| `protect` | number | No | `0.33` | 0 to 0.5 | Control how much of the original vocals' breath and voiceless consonants to leave in the AI vocals. Set 0.5 to disable |
| `output_format` | string | No | `mp3` | `mp3`, `wav` | wav for best quality and large file size, mp3 for decent quality and small file size |

## Pricing
- **Pay As You Go (PAYG) pricing**: you only pay for what you use.
- **Price**: $0.0043 per API unit.
- **1 API unit = 1 successful request with up to 17 seconds of processing.** A request that takes longer is billed one more unit for each started block of 17 seconds (for example, 20 seconds of processing is 2 units, $0.0086).
- Longer input audio and downloading a custom model take more processing time.
- **Example cost**: the example request above took 8.3 seconds of processing, so it used 1 API unit and cost $0.0043.
- **Failed or cancelled predictions are not billed.**
- **GET requests to check prediction status** are free and don't count toward API unit usage.
- **No monthly subscription fee.**

## API Documentation

### Base URL
All API requests go to: `https://prod.api.market/api/v1/magicapi/ai-speech-to-speech-conversion-pseudoram-rvc-v2`

### Authentication
Send your api.market API key in the `x-api-market-key` header:
```
x-api-market-key: YOUR_API_MARKET_KEY
```

### Asynchronous Processing
**IMPORTANT**: This API is asynchronous. The flow is:
1. Send a POST request to `/predictions`. A successful request returns **201 Created** with a prediction `id` and a `status` of `starting` or `processing`.
2. Poll GET `/predictions/{id}` with that ID.
3. Keep polling while `status` is `starting` or `processing`.
4. Stop when `status` is `succeeded` (read `output`), `failed` or `canceled` (read `error`).

A polling interval of 1 second and an overall timeout of 600 seconds (10 minutes) work well.

### Endpoints

#### Create a Prediction
```bash
curl -X POST "https://prod.api.market/api/v1/magicapi/ai-speech-to-speech-conversion-pseudoram-rvc-v2/predictions" \
  -H "Content-Type: application/json" \
  -H "x-api-market-key: YOUR_API_MARKET_KEY" \
  -d '{
    "version": "d18e2e0a6a6d3af183cc09622cebba8555ec9a9e66983261fc64c8b1572b7dce",
    "input": {
      "protect": 0.5,
      "f0_method": "rmvpe",
      "rvc_model": "CUSTOM",
      "index_rate": 1,
      "input_audio": "https://blog.api.market/wp-content/uploads/2026/10/rvc-v2-input-api-market.mp3",
      "pitch_change": 8,
      "rms_mix_rate": 1,
      "filter_radius": 1,
      "output_format": "mp3",
      "crepe_hop_length": 128,
      "custom_rvc_model_download_url": "https://huggingface.co/Argax/doofenshmirtz-RUS/resolve/main/doofenshmirtz.zip"
    }
  }'
```

Response (201 Created):
```json
{
  "id": "2v12vrqnbdrgc0d1ccwbn2r3k8",
  "version": "d18e2e0a6a6d3af183cc09622cebba8555ec9a9e66983261fc64c8b1572b7dce",
  "status": "starting",
  "created_at": "2026-10-22T10:47:19.798Z"
}
```

#### Check Prediction Status
```bash
curl -X GET "https://prod.api.market/api/v1/magicapi/ai-speech-to-speech-conversion-pseudoram-rvc-v2/predictions/{id}" \
  -H "x-api-market-key: YOUR_API_MARKET_KEY"
```

Response when the prediction has finished (200 OK):
```json
{
  "id": "2v12vrqnbdrgc0d1ccwbn2r3k8",
  "status": "succeeded",
  "output": "https://blog.api.market/wp-content/uploads/2026/10/rvc-v2-output-2v12vrqn-api-market-1.wav",
  "error": null,
  "created_at": "2026-10-22T10:47:19.798Z",
  "started_at": "2026-10-22T10:47:25.053Z",
  "completed_at": "2026-10-22T10:47:33.345Z"
}
```

`output` is a single URL string pointing to the converted audio file.

### Status Codes
- **201 Created**: prediction created (POST `/predictions`). This is the normal success response.
- **200 OK**: prediction details returned (GET `/predictions/{id}`).
- **400 Bad Request**: invalid input, for example an `rvc_model` that is not in the list or a `protect` value above 0.5.
- **401 Unauthorized**: missing or invalid API key.
- **404 Not Found**: no prediction with that ID.
- **429 Too Many Requests**: rate limit exceeded.
- **500 Internal Server Error** or **503 Service Unavailable**: temporary server problem; retry later.

## Rate Limits
All plans allow 10 requests per second. Requests above that return `429 Too Many Requests`; back off before retrying and honour any `Retry-After` header. Polling a prediction with GET is free but still counts toward the per-second limit, so poll every 1 to 5 seconds rather than in a tight loop.

## Error Handling
- Check the HTTP status code of every request and read the response body for the error message.
- A prediction can also end with `status` set to `failed`; the `error` field then explains why. Failed predictions are not billed.
- Make sure `input_audio` and `custom_rvc_model_download_url` are publicly reachable. The custom model URL must point to a downloadable RVC v2 model file, such as a .zip.
- Retry 429, 500 and 503 responses with exponential backoff. Fix the input before retrying a 400.

## Implementation Guide
### Example Code

#### JavaScript
```javascript
const apiKey = "YOUR_API_MARKET_KEY";
const baseUrl = "https://prod.api.market/api/v1/magicapi/ai-speech-to-speech-conversion-pseudoram-rvc-v2";

async function createAndWaitForPrediction(input) {
  const response = await fetch(`${baseUrl}/predictions`, {
    method: "POST",
    headers: {
      "Content-Type": "application/json",
      "x-api-market-key": apiKey
    },
    body: JSON.stringify({ version: "d18e2e0a6a6d3af183cc09622cebba8555ec9a9e66983261fc64c8b1572b7dce", input })
  });

  if (response.status !== 201) {
    throw new Error(`API error: ${response.status} ${await response.text()}`);
  }

  const prediction = await response.json();
  return pollForResults(prediction.id);
}

const delay = (ms) => new Promise((resolve) => setTimeout(resolve, ms));

async function pollForResults(predictionId) {
  const maxTimeout = 600000;
  const pollInterval = 1000;
  const startTime = Date.now();

  while (Date.now() - startTime < maxTimeout) {
    const response = await fetch(`${baseUrl}/predictions/${predictionId}`, {
      headers: { "x-api-market-key": apiKey }
    });

    if (!response.ok) {
      throw new Error(`API error: ${response.status} ${await response.text()}`);
    }

    const prediction = await response.json();

    if (prediction.status === "succeeded") {
      return prediction.output;
    }

    if (prediction.status === "failed" || prediction.status === "canceled") {
      throw new Error(`Prediction ${prediction.status}: ${prediction.error || "Unknown error"}`);
    }

    await delay(pollInterval);
  }

  throw new Error("Prediction timed out after 10 minutes");
}

createAndWaitForPrediction({
  protect: 0.5,
  f0_method: "rmvpe",
  rvc_model: "CUSTOM",
  index_rate: 1,
  input_audio: "https://blog.api.market/wp-content/uploads/2026/10/rvc-v2-input-api-market.mp3",
  pitch_change: 8,
  rms_mix_rate: 1,
  filter_radius: 1,
  output_format: "mp3",
  crepe_hop_length: 128,
  custom_rvc_model_download_url: "https://huggingface.co/Argax/doofenshmirtz-RUS/resolve/main/doofenshmirtz.zip"
}).then((output) => console.log(output));
```

#### Python
```python
import time
import requests

API_KEY = "YOUR_API_MARKET_KEY"
BASE_URL = "https://prod.api.market/api/v1/magicapi/ai-speech-to-speech-conversion-pseudoram-rvc-v2"


def create_and_wait_for_prediction(input_data):
    headers = {"Content-Type": "application/json", "x-api-market-key": API_KEY}
    response = requests.post(
        f"{BASE_URL}/predictions",
        headers=headers,
        json={"version": "d18e2e0a6a6d3af183cc09622cebba8555ec9a9e66983261fc64c8b1572b7dce", "input": input_data},
    )

    if response.status_code != 201:
        raise Exception(f"API error: {response.status_code} {response.text}")

    return poll_for_results(response.json()["id"])


def poll_for_results(prediction_id):
    max_timeout = 600
    poll_interval = 1
    headers = {"x-api-market-key": API_KEY}
    start_time = time.time()

    while time.time() - start_time < max_timeout:
        response = requests.get(f"{BASE_URL}/predictions/{prediction_id}", headers=headers)

        if response.status_code != 200:
            raise Exception(f"API error: {response.status_code} {response.text}")

        prediction = response.json()

        if prediction["status"] == "succeeded":
            return prediction["output"]

        if prediction["status"] in ("failed", "canceled"):
            raise Exception(f"Prediction {prediction['status']}: {prediction.get('error') or 'Unknown error'}")

        time.sleep(poll_interval)

    raise Exception("Prediction timed out after 10 minutes")


print(create_and_wait_for_prediction({
    "protect": 0.5,
    "f0_method": "rmvpe",
    "rvc_model": "CUSTOM",
    "index_rate": 1,
    "input_audio": "https://blog.api.market/wp-content/uploads/2026/10/rvc-v2-input-api-market.mp3",
    "pitch_change": 8,
    "rms_mix_rate": 1,
    "filter_radius": 1,
    "output_format": "mp3",
    "crepe_hop_length": 128,
    "custom_rvc_model_download_url": "https://huggingface.co/Argax/doofenshmirtz-RUS/resolve/main/doofenshmirtz.zip"
}))
```

## FAQ
- **How do I use my own voice model?**
  - Upload your RVC v2 model (for example a .zip) somewhere public and pass its URL as `custom_rvc_model_download_url`. It is downloaded and used regardless of `rvc_model`; the example also sets `rvc_model` to `CUSTOM`.
- **Can it convert singing as well as speech?**
  - Yes. The controls are written for vocals: shift the key with `pitch_change` and pick `rmvpe` for clarity or `mangio-crepe` for smoother vocals.
- **What do `index_rate`, `rms_mix_rate` and `protect` do?**
  - `index_rate` sets how much of the target voice's accent is used, `rms_mix_rate` blends the original loudness (0) with a fixed loudness (1), and `protect` keeps the original breath and voiceless consonants (0.5 turns this off).
- **Which output format do I get?**
  - Set `output_format` to `mp3` or `wav`. In the example, `mp3` was requested and the returned file was a WAV file, so check the extension of the `output` URL.
- **How is a request billed?**
  - $0.0043 per request for up to 17 seconds of processing, plus one more unit for each further started block of 17 seconds. The example took 8.3 seconds and cost $0.0043.
- **Is the 201 response from POST an error?**
  - No. 201 Created means the prediction was accepted. Use the returned `id` to poll GET `/predictions/{id}` until `status` is `succeeded`, `failed` or `canceled`.

## Contact Us
If you have any questions, reach out to us through the chat on the bottom right of the page. You can also email our support team at hello@magicapi.com.


Speech-to-speech voice conversion with RVC v2: load your own voice model from a URL, shift the pitch and get MP3 or WAV audio back.