相关方案
      社区内容

        # 超拟人交互 API

        # 前言

        # Web接口说明

        1. 必须符合WebSocket协议规范（RFC6455）。
        2. WebSocket握手成功后，用户在10秒内未发送请求数据，服务端会主动断开连接。
        3. 默认采用全双工交互模式，单个连接最长可持续30分钟。
        4. 服务端下发错误码后，客户端应重建连接。

        # 一、API介绍

        极速多模模拟人是端到端语音交互方案，支持音频输入与合成音频/视频流输出。

        # 1.1 开通授权

        1. 在xx平台应用管理页面领取/购买授权包
        2. 联系技术支持开通授权
        3. DeepSeek版本需额外申请授权

        # 二、API接入说明

        # 2.1 请求说明

        # 2.1.1请求地址

        wss://sparkos.xfym.cn/v1/openapi/chat

        # 2.1.2接口鉴权

        参考通用鉴权 (opens new window)

        # 2.1.3请求协议示例

        {
    "header": {
        "app_id": "", // 应用id
        "uid": "", // 用户唯一标识，用户关联用户交互历史，
        "status": 0, // 客户端发送数据的状态   0：开始；1：中间状态；2：结束
        "stmid": "1", // 交互轮数，客户端发起的交互计数。使用continuous_vad单工模式时，一次ws连接内的每一轮对话【必须】递增更新stmid。  使用continuous双工模式时，stmid固定不变。
        "scene": "sos_app", // 固定默认值
        "interact_mode":"continuous" //交互模式选择，支持continuous（双工）和 continuous_vad（单工）可选，默认continuous
    },
    "parameter": {
        "iat": {
            "iat": {
                "encoding": "utf8",
                "compress": "raw",
                "format": "json"
            },
            "vgap":50
        },
        "nlp": {
            "nlp": {
                "encoding": "utf8",
                "compress": "raw",
                "format": "json"
            },
            "new_session": "true",
            "personal":"人设 id",
            "prompt":"prompt 信息， 例如：你是小明，一个小学学生，热爱画画"
        },
        "tts": {
            "vcn": "x5_lingfeiyi_flow",
            "res_id": "xxxx",
            "res_gender":"",
            "speed": 50,
            "volume": 50,
            "pitch": 50,
            "tts": {
                "encoding": "raw",
                "sample_rate": 16000,
                "channels": 1,
                "bit_depth": 16,
                "frame_size": 0
            }
        },
        "avatar": {
            "avatar_id": "",
            "image": "",
            "encoding": "",
            "width": 512,
            "height": 512
        }
    },
    "payload": {
        "audio": {
            "status": 0,
            "audio": "base64的音频数据",
            "encoding": "raw",
            "sample_rate": 16000,
            "channels": 1,
            "bit_depth": 16,
            "frame_size": 0
        }
    }
}

        # 2.1.4请求参数

        # Header 参数
        参数名称 类型 必传 描述 取值限制/备注
        app_id string 是 应用ID maxLength:15
        uid string 是 授权的用户ID maxLength:64，需保证在appid下唯一
        status int 是 会话状态 0:第一帧，1:中间帧，2:最后一帧
        stmid string 是 会话ID/交互轮数 maxLength:32
        单工模式(continuous_vad)必须递增
        双工模式(continuous)固定不变
        需为可解析为整数的字符串，如"0","1"，具体可见下文示例
        scene string 是 情景模式 maxLength:16，从AIUI/飞云平台创建
        msc.lat float 否 纬度 [-90,90]
        msc.lng float 否 经度 [-180,180]
        interact_mode string 是 交互模式 continuous(全双工)
        continuous_vad(单工)
        bot_id string 否 指定bot ID maxLength:64
        示例：sos_app_deepseekv3
        # Parameter 参数
        # 语音识别(iat)
        参数名称 类型 必传 描述 取值限制
        iat.encoding string 是 结果编码 仅支持utf8
        iat.compress string 是 压缩类型 仅支持raw
        iat.format string 是 结果格式 仅支持json
        vgap int 否 静音断句阈值 默认80(800ms)
        范围40-1000(400-10000ms)
        # NLP(nlp)
        参数名称 类型 必传 描述 取值限制
        nlp.encoding string 是 结果编码 仅支持utf8
        nlp.compress string 是 压缩类型 仅支持raw
        nlp.format string 是 结果格式 仅支持json
        new_session string 否 是否新会话 "true"/"global":清空历史
        "false":保留历史
        prompt string 否 回复要求设定 通过该参数设定大模型回复风格、格式以及其他回答要求等
        # 语音合成(tts)
        参数名称 类型 必传 描述 取值限制
        tts.encoding string 是 输出音频编码 raw/lame/opus-wb/opus-swb
        默认raw:pcm音频
        lame:mp3格式音频
        sample_rate int 是 采样率 16000/24000
        默认16000Hz
        channels int 是 声道数 默认1(单声道)
        bit_depth int 是 位深 默认16bit
        vcn string 是 发音人 x5_lingxiaoyue_flow(聆小玥，女性助理)
        x5_lingfeiyi_flow(聆飞逸，男性助理)
        speed int 否 语速 0-100，默认50
        volume int 否 音量 0-100，默认50
        pitch int 否 音调 0-100，默认50
        # Payload 参数
        # 音频数据(audio)
        参数名称 类型 必传 描述 取值限制
        audio string 是 音频数据 base64编码
        单帧间隔≤40ms数据
        status int 是 音频状态 0:首帧，1:中间帧，2:末帧
        frame_size int 是 压缩帧大小 未压缩传0
        encoding string 是 编码格式 raw/opus
        默认raw
        sample_rate int 是 采样率 8000/16000
        默认16000Hz
        channels int 是 声道数 1/2
        默认1
        bit_depth int 是 位深 8/16
        默认16bit
        # 2.1.5 请求示例

        单工模式(continuous_vad)：端侧控制音频开始上传的模式，主要应用于嘈杂的环境，避免噪音导致误输入


        • 每轮对话【必须】更新header.stmid的值 并且不能重复以前的值。比如"0", "1", "2"...。
        • header.status： 会话的状态。第一个数据包的header.status是0，后续都是1，结束时传2。
        • payload.audio.status：音频的状态。每轮对话的音频，从0 1 ... 1 2。
        • 每轮对话的第一个数据包，发送全部的字段。后续的数据包，可以发送简化的内容（可以不发parameter部分，header中也只需要发送几个关键的字段）。

        ``

        建立ws连接
#【第一轮对话： header.stmid="0"】
# 第一轮对话的第一个数据包：需要发送全部的字段  (header.status=0, payload.audio.status=0)
{"header":{"app_id":"879230fc","uid":"thzhang4","status":0,"stmid":"0","scene":"ai-personality-1","os_sys":"android","interact_mode":"continuous_vad","pers_param":"{\"appid\":\"879230fc\",\"uid\":\"thzhang4\"}"},"parameter":{"iat":{"vgap":60,"dwa":"wpgs","iat":{"encoding":"utf8","compress":"raw","format":"json"},"eos":"800","domain":"sms"},"nlp":{"nlp":{"encoding":"utf8","compress":"raw","format":"json"},"new_session":"false"},"tts":{"vcn":"x5_lingxiaoyue_flow","speed":50,"volume":50,"pitch":50,"tts":{"encoding":"raw","sample_rate":16000,"channels":1,"bit_depth":16}}},"payload":{"audio":{"status":0,"audio":"base64的音频数据","encoding":"raw","sample_rate":16000,"channels":1,"bit_depth":16}}}
# 后续的数据包(header.status=1, payload.audio.status=1)
{"header":{"app_id":"879230fc","uid":"thzhang4","status":1,"stmid":"0","scene":"ai-personality-1"},"payload":{"audio":{"status":1,"audio":"base64的音频数据","encoding":"raw","sample_rate":16000,"channels":1,"bit_depth":16}}}
...
{"header":{"app_id":"879230fc","uid":"thzhang4","status":1,"stmid":"0","scene":"ai-personality-1"},"payload":{"audio":{"status":1,"audio":"base64的音频数据","encoding":"raw","sample_rate":16000,"channels":1,"bit_depth":16}}}
# 第一轮轮对话的最后一个数据包(header.status=1, payload.audio.status=2)
{"header":{"app_id":"879230fc","uid":"thzhang4","status":1,"stmid":"0","scene":"ai-personality-1"},"payload":{"audio":{"status":2,"audio":"base64的音频数据","encoding":"raw","sample_rate":16000,"channels":1,"bit_depth":16}}}


# 【第二轮对话  更新header.stmid="1"】
# 第二轮对话的第一个数据包：需要发送全部的字段  (header.status=1, payload.audio.status=0)
{"header":{"app_id":"879230fc","uid":"thzhang4","status":1,"stmid":"1","scene":"ai-personality-1","os_sys":"android","interact_mode":"continuous_vad","pers_param":"{\"appid\":\"879230fc\",\"uid\":\"thzhang4\"}"},"parameter":{"iat":{"vgap":60,"dwa":"wpgs","iat":{"encoding":"utf8","compress":"raw","format":"json"},"eos":"800","domain":"sms"},"nlp":{"nlp":{"encoding":"utf8","compress":"raw","format":"json"},"new_session":"false"},"tts":{"vcn":"x5_lingxiaoyue_flow","speed":50,"volume":50,"pitch":50,"tts":{"encoding":"raw","sample_rate":16000,"channels":1,"bit_depth":16}}},"payload":{"audio":{"status":0,"audio":"base64的音频数据","encoding":"raw","sample_rate":16000,"channels":1,"bit_depth":16}}}
# 后续的数据包(header.status=1, payload.audio.status=1)
{"header":{"app_id":"879230fc","uid":"thzhang4","status":1,"stmid":"1","scene":"ai-personality-1"},"payload":{"audio":{"status":1,"audio":"base64的音频数据","encoding":"raw","sample_rate":16000,"channels":1,"bit_depth":16}}}
...
{"header":{"app_id":"879230fc","uid":"thzhang4","status":1,"stmid":"1","scene":"ai-personality-1"},"payload":{"audio":{"status":1,"audio":"base64的音频数据","encoding":"raw","sample_rate":16000,"channels":1,"bit_depth":16}}}
# 第二轮轮对话的最后一个数据包(header.status=1, payload.audio.status=2)
{"header":{"app_id":"879230fc","uid":"thzhang4","status":1,"stmid":"1","scene":"ai-personality-1"},"payload":{"audio":{"status":2,"audio":"base64的音频数据","encoding":"raw","sample_rate":16000,"channels":1,"bit_depth":16}}}


# 【第三轮对话  更新header.stmid="2"】
# 第三轮对话的第一个数据包：需要发送全部的字段  (header.status=1, payload.audio.status=0)
# 后续的数据包(header.status=1, payload.audio.status=1)
# 第三轮轮对话的最后一个数据包(header.status=1, payload.audio.status=2)


# 结束会话
header.status=2
# 关闭ws连接
ws.close()

        双工模式(continuous)：应用于持续上传音频的模式，类似语音通话的场景


        • header.stmid的值【固定不变】
        • header.status： 会话的状态。第一个数据包的header.status是0，后续都是1，结束时传2。
        • payload.audio.status：音频的状态。第一个数据包的header.status是0，后续都是1，结束时传2。
        • 第一个数据包，发送全部的字段。后续的数据包，可以发送简化的内容（可以不发parameter部分，header中也只需要发送几个关键的字段）。
        • 【持续不断的发送音频】：即使用户没有说话，也要一直发送音频。服务侧会自动判断用户有没有说话。
        建立ws连接

# 第一个数据包：需要发送全部的字段  (header.status=0, payload.audio.status=0)
{"header":{"app_id":"879230fc","uid":"thzhang4","status":0,"stmid":"0","scene":"ai-personality-1","os_sys":"android","interact_mode":"continuous","test":"孙悟空","pers_param":"{\"appid\":\"879230fc\",\"uid\":\"thzhang4\"}"},"parameter":{"iat":{"vgap":60,"dwa":"wpgs","iat":{"encoding":"utf8","compress":"raw","format":"json"}},"nlp":{"nlp":{"encoding":"utf8","compress":"raw","format":"json"},"new_session":"false"},"tts":{"vcn":"x5_lingxiaoyue_flow","speed":50,"volume":50,"pitch":50,"tts":{"encoding":"raw","sample_rate":16000,"channels":1,"bit_depth":16}}},"payload":{"audio":{"status":0,"audio":"base64的音频","encoding":"raw","sample_rate":16000,"channels":1,"bit_depth":16}}}
# 后续的数据包(header.status=1, payload.audio.status=1)
{"header":{"app_id":"879230fc","uid":"thzhang4","status":1,"stmid":"0","scene":"ai-personality-1"},"payload":{"audio":{"status":1,"audio":"base64的音频数据","encoding":"raw","sample_rate":16000,"channels":1,"bit_depth":16}}}
...
{"header":{"app_id":"879230fc","uid":"thzhang4","status":1,"stmid":"0","scene":"ai-personality-1"},"payload":{"audio":{"status":1,"audio":"base64的音频数据","encoding":"raw","sample_rate":16000,"channels":1,"bit_depth":16}}}
...
# 即使用户没有说话，也要一直发送音频
...
# 结束会话 (header.status=2, payload.audio.status=2)
{"header":{"app_id":"879230fc","uid":"thzhang4","status":2,"stmid":"0","scene":"ai-personality-1"},"payload":{"audio":{"status":2,"audio":"base64的音频数据","encoding":"raw","sample_rate":16000,"channels":1,"bit_depth":16}}}

# 关闭ws连接
ws.close()

        # 三、响应说明

        # 3.1 响应示例

        首帧数据发送成功响应 // 不报错可以省略掉，通常用于返回参数错误、链接错误、鉴权等错误码

        {
    "header": {
        "code": 0,
        "message": "success",
        "sid": "xgo00010205@dx192743899eb0001822-audio-1",
        "status": 0,
        "stmid": "1"
    }
}

        正常结果示例

        {
    "header": {
        "code": 0,
        "message": "success",
        "sid": "xgo00010205@dx192743899eb0001822-audio-1",
        "status": 0,
        "stmid":"1"
    },
    "payload": {
        "event": {
            "compress": "",
            "encoding": "",
            "format": "",
            "seq": 0,
            "status": 0,
            "text": "eyJ0eXBlIjoiVmFkIiwiZGF0YSI6IiIsImtleSI6IkJvcyIsImRlc2MiOnt9fQ=="
        },
        "iat": {
            "compress": "raw",
            "encoding": "utf8",
            "format": "json",
            "seq": 1,
            "status": 2,
            "text": "eyJ0ZXh0Ijp7InNuIjoxLCJscyI6ZmFsc2UsImJnIjowLCJyZyI6bnVsbCwiZWQiOjAsInBncyI6IiIsInJzdCI6IiIsInNpZ24iOiIiLCJ3cyI6W3siYmciOjAsImN3IjpbeyJzYyI6MCwidyI6IuS9oCIsInBoIjoiIn1dfSx7ImJnIjowLCJjdyI6W3sic2MiOjAsInciOiLlj6siLCJwaCI6IiJ9XX0seyJiZyI6MCwiY3ciOlt7InNjIjowLCJ3Ijoi5LuA5LmIIiwicGgiOiIifV19LHsiYmciOjAsImN3IjpbeyJzYyI6MCwidyI6IuWQjeWtlyIsInBoIjoiIn1dfV19fQ=="
        },
         "nlp": {
            "compress": "",
            "encoding": "",
            "format": "",
            "seq": 0,
            "status": 0,
            "text": "5bGV56S65LqG5LiA5Liq"
        },
         "tts": {
            "compress": "",
            "encoding": "",
            "format": "",
            "seq": 0,
            "status": 0,
            "audio": "Base64 audio data"
        },
         "cbm_vms": {
            "compress": "",
            "encoding": "",
            "format": "",
            "seq": 0,
            "status": 0,
            "text": "Base64 vms connect/start/stop info"
        },
        
    }
}

        异常结果

        {
    "header": {
        "code": 10110,
        "message": "server licence error",
        "sid": "xgo00010205@dx192743899eb0001822-audio-1",
        "status": 2,
        "stmid": "1"
    }
}

        # 3.2 响应参数

        # 3.2.1协议字段说明

        # 字段

        字段名 类型 描述 备注
        header object -
        header.code int 服务错误码 0表示成功，其他值表示失败
        header.sid string 会话的sid
        header.status int 会话的状态 取值示例：0, 1, 1, ..., 1, 2
        header.stmid string 会话的id 长连接中每个会话的唯一ID；客户端发起请求后云端迭代返回（如0-1→0-2→0-3）
        payload object 数据字段 包含详细业务数据
        payload.event object 事件数据
        payload.event.text base64 数据内容 云端的识别会判断用户音频的断点信息，会下发相应的事件，目前主要分为三类事件，包括开始说话，结束说话，无人说话
        payload.event.text.status int 数据的状态 可选值0，1，2。 其中0表示首帧结果，1表示中间帧，2表示最后一帧。
        payload.event.text.encoding string 数据的编码格式 --
        payload.event.text.seq int 数据段的编号 --
        payload.event.text.format string 数据内容的格式 --
        payload.event.text.compress string 数据内容的压缩方法 --
        payload.iat object 用户语音输入的识别结果 详见下方参数示例
        payload.nlp object 模型回复的文本内容 详见下方参数示例
        payload.tts object 模型合成的音频数据 详见下方参数示例

        # 3.2.2 数据字段详解

        事件数据

        云端的识别会判断用户音频的断点信息，会下发相应的事件，目前主要分为三类事件，包括开始说话，结束说话，无人说话，
事件结果在 payload.event.text 中

        事件目前分三种事件，详细说明如下：

        // 以下结果已经 base64decode text的内容
Bos 事件，检测到音频中有人说话，触发该事件返回
{"type":"Vad","data":"","key":"Bos","desc":{}}

Eos事件，检测到音频中子句说完了，触发该事件返回
{"type":"Vad","data":"","key":"Eos","desc":{}}

Silence 事件，检测到音频无人说话，最终结束事件，收到后结束会话即可，一般静音事件过长，或者上传status=2时会触发
{"type":"Vad","data":"","key":"Silence","desc":{}}

        识别数据

        识别结果在 payload.iat.text 字段中（注意示例为流式结果，当前默认识别流式结果）

         //以下结果已经 base64decode text的内容
{"sn":1,"ls":false,"bg":0,"ed":0,"pgs":"apd","ws":[{"bg":0,"cw":[{"sc":0.00,"w":"给"}]}]}
{"sn":2,"ls":false,"bg":0,"ed":0,"pgs":"rpl","rg":[1,1],"ws":[{"bg":0,"cw":[{"sc":0.00,"w":"给"}]},{"bg":0,"cw":[{"sc":0.00,"w":"1385"}]}]}
{"sn":3,"ls":false,"bg":0,"ed":0,"pgs":"rpl","rg":[1,2],"ws":[{"bg":0,"cw":[{"sc":0.00,"w":"给"}]},{"bg":0,"cw":[{"sc":0.00,"w":"13856"}]}]}
{"sn":4,"ls":false,"bg":0,"ed":0,"pgs":"rpl","rg":[1,3],"ws":[{"bg":0,"cw":[{"sc":0.00,"w":"给"}]},{"bg":0,"cw":[{"sc":0.00,"w":"138569"}]}]}
{"sn":5,"ls":false,"bg":0,"ed":0,"pgs":"rpl","rg":[1,4],"ws":[{"bg":0,"cw":[{"sc":0.00,"w":"给"}]},{"bg":0,"cw":[{"sc":0.00,"w":"1385690"}]}]}
{"sn":6,"ls":false,"bg":0,"ed":0,"pgs":"rpl","rg":[1,5],"ws":[{"bg":0,"cw":[{"sc":0.00,"w":"给"}]},{"bg":0,"cw":[{"sc":0.00,"w":"13856901"}]}]}
{"sn":7,"ls":false,"bg":0,"ed":0,"pgs":"rpl","rg":[1,6],"ws":[{"bg":0,"cw":[{"sc":0.00,"w":"给"}]},{"bg":0,"cw":[{"sc":0.00,"w":"138569012"}]}]}
{"sn":8,"ls":false,"bg":0,"ed":0,"pgs":"rpl","rg":[1,7],"ws":[{"bg":0,"cw":[{"sc":0.00,"w":"给"}]},{"bg":0,"cw":[{"sc":0.00,"w":"1385690123"}]}]}
{"sn":9,"ls":false,"bg":0,"ed":0,"pgs":"rpl","rg":[1,8],"ws":[{"bg":0,"cw":[{"sc":0.00,"w":"给"}]},{"bg":0,"cw":[{"sc":0.00,"w":"13856901234"}]}]}
{"sn":10,"ls":false,"bg":0,"ed":0,"pgs":"rpl","rg":[1,9],"ws":[{"bg":0,"cw":[{"sc":0.00,"w":"给"}]},{"bg":0,"cw":[{"sc":0.00,"w":"13856901234"}]},{"bg":0,"cw":[{"sc":0.00,"w":"充"}]}]}
{"sn":11,"ls":false,"bg":0,"ed":0,"pgs":"rpl","rg":[1,10],"ws":[{"bg":0,"cw":[{"sc":0.00,"w":"给"}]},{"bg":0,"cw":[{"sc":0.00,"w":"13856901234"}]},{"bg":0,"cw":[{"sc":0.00,"w":"重"}]},{"bg":0,"cw":[{"sc":0.00,"w":"六"}]}]}
{"sn":12,"ls":false,"bg":0,"ed":0,"pgs":"rpl","rg":[1,11],"ws":[{"bg":0,"cw":[{"sc":0.00,"w":"给"}]},{"bg":0,"cw":[{"sc":0.00,"w":"13856901234"}]},{"bg":0,"cw":[{"sc":0.00,"w":"充"}]},{"bg":0,"cw":[{"sc":0.00,"w":"六"}]}]}
{"sn":13,"ls":false,"bg":0,"ed":0,"pgs":"rpl","rg":[1,12],"ws":[{"bg":0,"cw":[{"sc":0.00,"w":"给"}]},{"bg":0,"cw":[{"sc":0.00,"w":"13856901234"}]},{"bg":0,"cw":[{"sc":0.00,"w":"充"}]},{"bg":0,"cw":[{"sc":0.00,"w":"六"}]},{"bg":0,"cw":[{"sc":0.00,"w":"千"}]}]}
{"sn":14,"ls":false,"bg":0,"ed":0,"pgs":"rpl","rg":[1,13],"ws":[{"bg":0,"cw":[{"sc":0.00,"w":"给"}]},{"bg":0,"cw":[{"sc":0.00,"w":"13856901234"}]},{"bg":0,"cw":[{"sc":0.00,"w":"充"}]},{"bg":0,"cw":[{"sc":0.00,"w":"6800"}]}]}
{"sn":15,"ls":false,"bg":0,"ed":0,"pgs":"rpl","rg":[1,14],"ws":[{"bg":0,"cw":[{"sc":0.00,"w":"给"}]},{"bg":0,"cw":[{"sc":0.00,"w":"13856901234"}]},{"bg":0,"cw":[{"sc":0.00,"w":"充"}]},{"bg":0,"cw":[{"sc":0.00,"w":"6880"}]}]}
{"sn":16,"ls":false,"bg":0,"ed":0,"pgs":"rpl","rg":[1,15],"ws":[{"bg":0,"cw":[{"sc":0.00,"w":"给"}]},{"bg":0,"cw":[{"sc":0.00,"w":"13856901234"}]},{"bg":0,"cw":[{"sc":0.00,"w":"充"}]},{"bg":0,"cw":[{"sc":0.00,"w":"6888"}]}]}
{"sn":17,"ls":false,"bg":0,"ed":0,"pgs":"rpl","rg":[1,16],"ws":[{"bg":0,"cw":[{"sc":0.00,"w":"给"}]},{"bg":0,"cw":[{"sc":0.00,"w":"13856901234"}]},{"bg":0,"cw":[{"sc":0.00,"w":"充"}]},{"bg":0,"cw":[{"sc":0.00,"w":"6888"}]},{"bg":0,"cw":[{"sc":0.00,"w":"点"}]}]}
{"sn":18,"ls":false,"bg":0,"ed":0,"pgs":"rpl","rg":[1,17],"ws":[{"bg":0,"cw":[{"sc":0.00,"w":"给"}]},{"bg":0,"cw":[{"sc":0.00,"w":"13856901234"}]},{"bg":0,"cw":[{"sc":0.00,"w":"充"}]},{"bg":0,"cw":[{"sc":0.00,"w":"6888"}]},{"bg":0,"cw":[{"sc":0.00,"w":"."}]},{"bg":0,"cw":[{"sc":0.00,"w":"8"}]}]}
{"sn":19,"ls":false,"bg":0,"ed":0,"pgs":"rpl","rg":[1,18],"ws":[{"bg":0,"cw":[{"sc":0.00,"w":"给"}]},{"bg":0,"cw":[{"sc":0.00,"w":"13856901234"}]},{"bg":0,"cw":[{"sc":0.00,"w":"充"}]},{"bg":0,"cw":[{"sc":0.00,"w":"6888"}]},{"bg":0,"cw":[{"sc":0.00,"w":"."}]},{"bg":0,"cw":[{"sc":0.00,"w":"8"}]},{"bg":0,"cw":[{"sc":0.00,"w":"元"}]}]}
{"sn":20,"ls":false,"bg":0,"ed":0,"pgs":"rpl","rg":[1,19],"ws":[{"bg":0,"cw":[{"sc":0.00,"w":"给"}]},{"bg":0,"cw":[{"sc":0.00,"w":"13856901234"}]},{"bg":0,"cw":[{"sc":0.00,"w":"充"}]},{"bg":0,"cw":[{"sc":0.00,"w":"6888"}]},{"bg":0,"cw":[{"sc":0.00,"w":"."}]},{"bg":0,"cw":[{"sc":0.00,"w":"8"}]},{"bg":0,"cw":[{"sc":0.00,"w":"元"}]},{"bg":0,"cw":[{"sc":0.00,"w":"话费"}]}]}
{"sn":21,"ls":false,"bg":0,"ed":0,"pgs":"rpl","rg":[1,20],"ws":[{"bg":0,"cw":[{"sc":0.00,"w":"给"}]},{"bg":0,"cw":[{"sc":0.00,"w":"13856901234"}]},{"bg":0,"cw":[{"sc":0.00,"w":"充"}]},{"bg":0,"cw":[{"sc":0.00,"w":"6888"}]},{"bg":0,"cw":[{"sc":0.00,"w":"."}]},{"bg":0,"cw":[{"sc":0.00,"w":"8"}]},{"bg":0,"cw":[{"sc":0.00,"w":"元"}]},{"bg":0,"cw":[{"sc":0.00,"w":"话费"}]}]}
{"sn":22,"ls":true,"bg":0,"ed":0,"pgs":"apd","ws":[{"bg":0,"cw":[{"sc":0.00,"w":"。"}]}]}

        字段含义：

        Json字段 类型 描述
        pgs string 开启wpgs会有此字段 取值为 "apd"时表示该片结果是追加到前面的最终结果；取值为"rpl" 时表示替换前面的部分结果，替换范围为rg字段
        rg array 替换范围，开启wpgs会有此字段 假设值为[2,5]，则代表要替换的是第2次到第5次返回的结果

        **后台保证返回pgs结果无交叉嵌套。根据参数说明得到上述示例的最终听写结果为 “给13856901234充6888.8元话费。” **

        解析规则：

        1. 普通结果，每个识别结果均是独立结果，直接解析即可
        2. PGS结果，需要解析BOS和EOS之前的所有结果进行拼接，最后的结果一般在rst=rlt中

        回复文本数据

        模型结果在 payload.nlp.text 结果中，流式下发

        //以下结果已经 base64decode text的内容
今天
天气晴朗
，气温 10～12℃，
东北风微风

        合成音频数据

        合成音频数据在 payload.tts.audio 中，需要客户端 base64 decode，可按照音频参数encoding、sample_rate、bit_depth 播放

        合成音频数据示例如下，音频数据段大，流式下发：

        {
    "payload": {
        "tts": {
            "audio": "Base64 data ",
            "bit_depth": 16,
            "channels": 1,
            "encoding": "raw",
            "frame_size": 0,
            "sample_rate": 24000,
            "seq": 1,
            "status": 0
        }
    }
}

        # 四、示例代码

        Python示例demo (opens new window)

        Java示例demo (opens new window)

        在这篇文章中：
        在线咨询
        体验中心