---
title: Marco_Longspeech
canonical_url: "https://www.modelscope.ai/datasets/ATH-MaaS/Marco_Longspeech"
md_url: "https://www.modelscope.ai/datasets/ATH-MaaS/Marco_Longspeech.md"
repository: ATH-MaaS/Marco_Longspeech
last_updated: 2026-07-14
license: apache-2.0
storage_size: "2.5 GB"
downloads: 94
stars: 0
---

# Marco_Longspeech

> Marco_Longspeech - An open-source dataset by ATH-MaaS on ModelScope. LongSpeech 是一个多任务长语音理解数据集，包含 8 种不同的语音理解任务。

ATH-MaaS/Marco_Longspeech is a dataset on ModelScope. totalling 2.5 GB. licensed under apache-2.0.

- **Repository**: ATH-MaaS/Marco_Longspeech
- **License**: apache-2.0
- **Storage size**: 2.5 GB
- **Downloads**: 94
- **Stars**: 0
- **Last updated**: 2026-07-14

Source: https://www.modelscope.ai/datasets/ATH-MaaS/Marco_Longspeech

---

# LongSpeech Dataset

LongSpeech 是一个多任务长语音理解数据集，包含 8 种不同的语音理解任务。

## 📊 数据集统计

### 任务统计

| Task | Train | Val | Test | Total | Unique Audios |
|------|-------|-----|------|-------|---------------|
| ASR | 71,275 | 15,273 | 15,274 | 101,822 | 101,822 |
| Temporal_Relative_QA | 5,886 | 1,261 | 1,262 | 8,409 | 8,409 |
| summary | 4,366 | 935 | 937 | 6,238 | 6,238 |
| content_separation | 5,887 | 1,261 | 1,263 | 8,411 | 8,411 |
| emotionQA | 5,887 | 1,261 | 1,263 | 8,411 | 8,411 |
| speaker_count | 5,887 | 1,261 | 1,263 | 8,411 | 8,411 |
| translation | 29,435 | 6,307 | 6,309 | 42,051 | 8,411 |
| language_detection | 14,789 | 3,169 | 3,170 | 21,128 | 21,128 |
| **Total** | **143,412** | **30,728** | **30,741** | **204,881** | - |


### 音频子集统计

| Subset | WAV Files | all_audios.jsonl | metadata.json |
|--------|-----------|------------------|----------------|
| LongSpeech_p1 | 29,539 | ✓ | ✓ |
| LongSpeech_p2 | 22,107 | ✓ | ✓ |
| LongSpeech_p3 | 50,176 | ✓ | ✓ |
| **Total** | **101,822** | - | - |


## 📁 数据集结构

```
LongSpeech-Dataset/
├── LongSpeechQA/                    # 8个任务的问答数据
│   ├── ASR/                         # 语音识别任务
│   │   ├── train.jsonl
│   │   ├── val.jsonl
│   │   └── test.jsonl
│   ├── Temporal_Relative_QA/        # 时序相关问答
│   ├── summary/                     # 摘要任务
│   ├── content_separation/          # 内容分离任务
│   ├── emotionQA/                   # 情感问答任务
│   ├── speaker_count/               # 说话人计数任务
│   ├── translation/                 # 翻译任务
│   └── language_detection/          # 语言检测任务
├── LongSpeech_p1/
│   ├── wavs/
│   ├── all_audios.jsonl
│   └── metadata.json
├── LongSpeech_p2/
│   ├── wavs/
│   ├── all_audios.jsonl
│   └── metadata.json
├── LongSpeech_p3/
│   ├── wavs/
│   ├── all_audios.jsonl
│   └── metadata.json
└── README.md
```

## 🎯 任务说明

| 任务 | 说明 |
|------|------|
| **ASR** | 自动语音识别，将语音转换为文本 |
| **Temporal_Relative_QA** | 时序相关问答，理解语音中的时间关系 |
| **summary** | 语音摘要，生成语音内容的简洁摘要 |
| **content_separation** | 内容分离，区分和提取不同的语音内容 |
| **emotionQA** | 情感问答，识别和分析语音中的情感 |
| **speaker_count** | 说话人计数，统计语音中的说话人数量 |
| **translation** | 语音翻译，将语音翻译为目标语言文本 |
| **language_detection** | 语言检测，识别语音所使用的语言 |

## 📝 数据格式

每个任务的 jsonl 文件包含以下格式的数据：

```json
{
    "language": "en",
    "task": "ASR",
    "messages": [
        {
            "role": "user",
            "audio": "LongSpeech_p1/wavs/013429.wav",
            "content": "Detect the language and recognize the speech: <|en|>"
        },
        {
            "role": "assistant",
            "content": "We wont feel compelled in any way to pay at the top end or...."
        }
    ]
}
```

### 字段说明

- `language`: 语音的语言代码 (en/zh 等)
- `task`: 任务类型
- `messages`: 对话消息列表
  - `role`: 角色 (user/assistant)
  - `audio`: 音频文件的相对路径
  - `content`: 文本内容（用户指令或助手回复）

## 🚀 使用方法

### 使用 Hugging Face Datasets 加载

```python
from datasets import load_dataset

# 加载特定任务的数据
dataset = load_dataset("your-username/LongSpeech-Dataset", data_files={
    "train": "LongSpeechQA/ASR/train.jsonl",
    "val": "LongSpeechQA/ASR/val.jsonl",
    "test": "LongSpeechQA/ASR/test.jsonl"
})

print(dataset)
```

### 加载音频文件

```python
import os
from datasets import load_dataset, Audio

# 假设数据集已下载到本地
dataset = load_dataset("json", data_files="LongSpeechQA/ASR/train.jsonl")

# 获取音频路径
for example in dataset["train"]:
    audio_path = example["messages"][0].get("audio")
    if audio_path:
        # 根据实际下载路径调整
        full_path = os.path.join("your_download_path", audio_path)
        print(f"Audio: {full_path}")
```