Retell is generating assistant text, but no audio ...
# ❓・ai-evy-supportm
Marlon
03/29/2026, 7:59 AM@AI Evy Retell is generating assistant text, but no audio is audible.
What I verified:
- In Test Audio, I can see the assistant text appear on screen
- I hear other tabs and system audio fine
- browser tab is not muted
- I changed voices and still no audio
- I tried another browser path and still no audio
- on phone calls, I hear the recording disclaimer, then silence
- this is affecting audio playback/TTS, not transcript generation
Please check voice playback/TTS for this workspace or agent path.
Can you confirm whether TTS audio is actually being generated for this agent/workspace, or whether only transcript text is being produced?
https://cdn.discordapp.com/attachments/1302816169367044149/1487722845885829240/content.png?ex=69ca2d5f&is=69c8dbdf&hm=2a3ceb8042831e92d27e6e4563276054961f714dd2f2952e66c680081e7ea2f6&
a
AI Evy
03/29/2026, 7:59 AMThanks for tagging me @Marlon. I'll look through relevant sources and get back to you shortly.
AI Evy
03/29/2026, 7:59 AM### Answer
The issue you're experiencing—where transcripts appear but no TTS audio plays—isn't directly covered in the available sources. However, I can confirm that Retell AI's architecture includes a distinct TTS (text-to-speech) component that synthesizes audio from response text [(1) ]() .
Based on the documentation, here are the key TTS-related configurations to verify in your agent settings [(2) ]() :
Voice Configuration:
- `voice_id`: Ensure a valid voice ID is set (e.g., "11labs-Adrian")
- `voice_model`: Verify the voice model is properly configured for your selected provider
- `fallback_voice_ids`: Check if fallback voices are configured correctly
Audio Settings:
- `volume`: Confirm volume is set appropriately (default is 1)
- `voice_speed`: Verify speed settings aren't causing issues
- `voice_temperature`: Check temperature configuration
TTS Provider Status:
Retell AI integrates with multiple TTS providers including ElevenLabs, Cartesia, MiniMax, and OpenAI [(3) ]() . The system should stream TTS output for lower latency [(1) ]() .
Since you've already verified browser audio, changed voices, and tested multiple browsers, this suggests a potential workspace or agent configuration issue rather than a client-side problem.
For workspace-level diagnostics and confirmation of whether TTS audio is actually being generated server-side, you'll need direct support:
- Visit [Contact Sales]()
- Ask questions on [Discord]()
- Visit
Would you like me to provide a general answer based on common TTS troubleshooting approaches?
AI Evy
03/29/2026, 7:59 AM
To learn more, see the sources I considered:
[[1] How to Build A Good AI Voice Agent | Retell AI | Retell AI](https://www.retellai.com/blog/how-to-build-a-good-voice-agent)
[[2] Create Voice Agent](https://docs.retellai.com/api-references/create-agent)
[[3] Set language for your agent](https://docs.retellai.com/agent/language)
AI Evy
03/29/2026, 7:59 AMFeel free to tag @AI Evy with additional questions. I also appreciate feedback, it helps me improve.
m
Marlon
03/29/2026, 8:01 AM@AI Evy I already changed voices, browsers, and local audio output.
Please answer these exact questions for my workspace/agent:
1. Is TTS audio actually being generated server-side for these test runs, yes or no?
2. If yes, what provider/model is being used in the failing runs?
3. If no, what exact error or failure reason do you see on your side?
4. Can you see whether the agent is producing transcript text only, without successful audio synthesis/playback?
5. Is this a workspace-level TTS issue on your side?
The key symptom is:
- text appears in Retell Test Audio
- no sound is audible
- phone calls play disclaimer, then silence
Please investigate the failing call/test logs directly rather than sending general TTS settings.
a
AI Evy
03/29/2026, 8:01 AMThanks for the follow-up, @Marlon. I'll get back to you soon.
AI Evy
03/29/2026, 8:01 AMFeel free to tag @AI Evy with additional questions. I also appreciate feedback, it helps me improve.
AI Evy
03/29/2026, 8:01 AM### Answer
I don't have access to your specific workspace logs, call IDs, or server-side diagnostic data to answer your questions about whether TTS audio is being generated for your specific test runs.
The available sources don't provide workspace-level diagnostic capabilities or the ability to investigate individual call logs directly. To get answers to your specific questions about server-side TTS generation, provider/model usage, error logs, and workspace-level issues, you need direct support access:
- Visit [Contact Sales]()
- Ask questions on [Discord]()
- Visit
When contacting support, provide specific call IDs from your failing test runs so they can investigate the server-side logs and determine whether TTS audio is being generated but not delivered, or if synthesis is failing entirely.
AI Evy
03/29/2026, 8:01 AMmessage has been deleted