[Feature Request] Support sending voice messages (VOICE type) to WeChat users

[Jefsky]

Problem

When the AI sends an audio file (e.g., TTS output) to a WeChat user via openclaw-weixin, the file is sent as a generic file attachment (FILE type). Users receive a .mp3/.wav file rather than a playable WeChat voice message.

Expected Behavior

The plugin should support sending voice messages using WeChat's native VOICE message type (MessageItemType.VOICE: 3), so users receive a voice message they can play inline, identical to how humans send voice notes.

Plugin Version

• Plugin: @tencent-weixin/openclaw-weixin v2.1.3
• OpenClaw: 2026.4.1+

Analysis

The plugin's send-media.ts already handles:

• video/* → sendVideoMessageWeixin + UploadMediaType.VIDEO: 2
• image/* → sendImageMessageWeixin + UploadMediaType.IMAGE: 1
• generic files → sendFileMessageWeixin + UploadMediaType.FILE: 3

Audio files (audio/*) currently fall through to the generic file handler. A sendVoiceMessageWeixin function and corresponding upload path using UploadMediaType.VOICE: 4 would need to be added in send-media.ts, similar to the existing video/image patterns.

Additional Context

The WeChat API supports VOICE message type (MessageItemType.VOICE: 3). The upstream CDN upload and message sending infrastructure already exists in the plugin — only the voice-specific routing and VoiceItem construction are missing.

[morece]

I encountered this issue as well. When can the official fix it?

[cgf120]

#9 已经提了 作者没合并

[morece]

期待 发自我的iPhone

…

------------------ Original ------------------ From: CuiGaofeng ***@***.***> Date: Fri,Apr 3,2026 11:20 AM To: Tencent/openclaw-weixin ***@***.***> Cc: morece ***@***.***>, Comment ***@***.***> Subject: Re: [Tencent/openclaw-weixin] [Feature Request] Support sending voice messages (VOICE type) to WeChat users (Issue #18) cgf120 left a comment (Tencent/openclaw-weixin#18) #9 已经提了 作者没合并 — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you commented.Message ID: ***@***.***>

[morece]

有分支吗？可以自己编译吗 发自我的iPhone

…

------------------ Original ------------------ From: CuiGaofeng ***@***.***> Date: Fri,Apr 3,2026 11:20 AM To: Tencent/openclaw-weixin ***@***.***> Cc: morece ***@***.***>, Comment ***@***.***> Subject: Re: [Tencent/openclaw-weixin] [Feature Request] Support sending voice messages (VOICE type) to WeChat users (Issue #18) cgf120 left a comment (Tencent/openclaw-weixin#18) #9 已经提了 作者没合并 — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you commented.Message ID: ***@***.***>

[Lrs-GreenBottle]

Still reproducible here on:

• OpenClaw 2026.4.2
• openclaw-weixin 2.1.6
• channel: openclaw-weixin

Repro:

1. Use OpenClaw TTS on the Weixin channel
2. Receive the outbound audio on Weixin

Actual behavior:

• the recipient gets an mp3 file attachment
• not a native Weixin voice message bubble

Expected behavior:

• when the outbound audio is in a supported format, send it as a native Weixin voice message instead of a generic file

This is especially noticeable for TTS testing, because users naturally expect a voice message rather than a downloadable mp3.

Looks related to PR #9, which appears to implement the missing native voice send path.

[Lrs-GreenBottle]

Additional evidence after local patch testing:

Environment:

• OpenClaw 2026.4.5
• openclaw-weixin 2.1.6
• direct chat on openclaw-weixin

What I tested:

1. Added a local outbound VOICE send path (equivalent to PR Add outbound Weixin native voice-message support #9)
2. Restarted Gateway and verified the patched code was loaded
3. Tested both:
   • .mp3 with encode_type = 7
   • .silk with encode_type = 6

Observed behavior:

• Upload succeeds
• sendVoiceMessageWeixin returns success
• outbound logs show the voice path is actually used
• but the WeChat client receives nothing at all (not even a broken/empty voice bubble)

Relevant logs:

• sendWeixinMediaFile: uploading voice ... encodeType=7
• sendVoiceMessageWeixin: success ...
• sendWeixinMediaFile: uploading voice ... encodeType=6
• sendVoiceMessageWeixin: success ...

This suggests the problem is deeper than just “missing outbound voice routing”.

Possible conclusion:

• either downstream bot delivery does not actually support outbound native VOICE messages yet
• or the current public protocol/docs are still missing required fields/constraints for VOICE

One more doc inconsistency I noticed:

• the README capability table for sendMessage still says only text/image/video/file
• getUploadUrl docs only document media_type values 1=IMAGE, 2=VIDEO, 3=FILE
• but the message schema later still exposes voice_item

So at least from an integrator perspective, current docs + protocol shape strongly imply VOICE is incomplete or not actually supported for outbound bot delivery.

[self-code-j]

四月份提的，六月份都没希望吗，我想给我的小助理添加语音功能。我的小助理可以发语音16 kHz 单声道 Ogg Speex，也能创建语音气泡，但客户端无法正常解码播放，什么时候能修啊