This guide covers common issues you might encounter while using PeachDesk Voice Studio, along with solutions. Voice Studio is a hosted web app inside the PeachDesk dashboard — there’s nothing to install, and all synthesis runs on Bibha’s GPU infrastructure. Most problems come down to your account, your browser, or your input audio rather than anything you need to fix locally.

​
Access & Sign-in Issues

​
Voice Studio Won’t Open

Symptoms:
  • Clicking Voice Studio in the PeachDesk dashboard does nothing
  • The Studio loads to a blank screen
Solutions:
  • Refresh the page and try again.
  • Make sure you’re signed in to PeachDesk and that your session hasn’t expired — sign out and back in if needed.
  • Confirm your workspace has Voice Studio enabled. If you don’t see the tab at all, ask your workspace admin to check your plan and permissions.
  • Try a different browser or an incognito/private window to rule out a stale extension or cache.
Voice Studio targets recent versions of Chrome, Edge, Safari, and Firefox. Very old browsers may not render the audio player or waveform correctly.

​
”Session Expired” or Sudden Sign-out

Solutions:
  • Sign back in from the PeachDesk dashboard. Your voice profiles, takes, and history are stored server-side and will still be there.
  • If you’re repeatedly signed out, disable browser extensions that block cookies or local storage for app.peachdeskai.com.

​
Generation Issues

​
Generation Is Slow or Stuck

Symptoms:
  • A request sits on “Generating…” for a long time
  • The progress indicator doesn’t advance
Explanation: Synthesis runs on PeachDesk’s shared GPU infrastructure. The first request after a quiet period may take a little longer while capacity for the selected model is warmed up; subsequent requests are much faster. Solutions:
Most slow generations are transient warm-up or queueing delays. Give it a few seconds, and if it doesn’t complete, click Generate again.
For very long passages, split the text into smaller segments and generate them separately. Shorter requests complete faster and are easier to re-run if you want to tweak one part.
Different engines have different speed/quality trade-offs. If you don’t need the heaviest model, pick a lighter one from the model dropdown in the Generation tab for faster turnaround.
A flaky network can stall the request or the audio download. Confirm app.peachdeskai.com is reachable and not blocked by a corporate firewall or proxy.

​
Generation Fails with an Error

Symptoms:
  • A red error appears after clicking Generate
  • “Generation failed” or “Something went wrong”
Solutions:
  • Retry — most failures are transient.
  • Make sure a voice and a model are both selected and that the text box isn’t empty.
  • Remove unusual characters or very long unbroken strings (URLs, code, long numbers) that the model may struggle to parse.
  • If a specific model consistently fails, switch to another engine from the dropdown and report the issue (see Still Having Issues?).

​
Poor Voice Quality

Symptoms:
  • Robotic or unnatural voice
  • Missing emotion or prosody
  • Pronunciation errors
Solutions:
1

Improve your cloning samples

If you’re using a Clone from audio voice profile, the output mirrors your reference audio. Use 10–30 seconds of clear speech, avoid background noise, and keep a consistent speaking tone. Adding multiple clean samples from the same speaker in the Voices tab improves results.
2

Match the speaking style

The generated voice mimics the tone and energy of your samples. A monotone reference produces a monotone result — record with the delivery you want to hear back.
3

Adjust your text formatting

Use proper punctuation, add commas for natural pauses, and capitalize proper nouns. Well-formatted text gives the model the cues it needs for natural prosody.
4

Try a different model or voice

Some engines and Built-in voices handle certain languages, accents, or styles better than others. Experiment with the model dropdown to find the best fit for your text.
For tricky pronunciations, try spelling the word phonetically in the text (for example, write a name the way it sounds). This often resolves mispronunciations without any other change.

​
Audio Issues

​
No Audio Playback

Symptoms:
  • The generated take won’t play
  • The play button doesn’t respond
Solutions:
  • Check your system audio settings and make sure an output device is connected and not muted.
  • Confirm the browser tab isn’t muted (most browsers show a mute icon on the tab).
  • Download the take and play it in a separate media player to confirm the audio itself is fine.
  • Refresh the page if the in-browser player becomes unresponsive.

​
Crackling or Distorted Audio

Symptoms:
  • Static, distortion, or clipping in the output
Solutions:
  • Check whether your Clone from audio samples already contain distortion or clipping — the model can reproduce it. Re-record or re-upload cleaner samples in the Voices tab.
  • Re-generate the take; an occasional artifact can be a one-off.
  • Lower your local playback volume if the distortion only appears at high volume.

​
Effects & Takes

​
An Effect Didn’t Apply

Symptoms:
  • A take looks unchanged after applying an effect
  • The new version isn’t visible
Explanation: Effects in Voice Studio are applied as versioned takes — applying an effect creates a new take rather than overwriting the original. Solutions:
  • Look for the newly created version in the take’s history rather than expecting the original to change.
  • If the new version is missing, refresh and re-apply the effect from the Effects tab.
  • Make sure you selected a source take before applying the effect.

​
Voice Profile Issues

​
Voice Clone Upload Fails

Symptoms:
  • “Upload failed” when adding a sample under Clone from audio
  • The sample never finishes processing
Solutions:
  • Use a supported audio format (such as WAV, MP3, or M4A) and keep individual samples reasonably short (10–30 seconds works well).
  • Check that the file isn’t corrupted by playing it locally first.
  • Retry on a stable connection — large uploads can fail on flaky networks.
  • If processing stalls, remove the sample and re-add it.

​
A Voice Profile Is Missing

Solutions:
  • Confirm you’re signed in to the correct PeachDesk workspace — profiles are scoped to your workspace.
  • Refresh the Voices tab; profiles are stored server-side and should reappear.
  • If a profile was created by a teammate, check that it was shared with you or your workspace.

​
History & Saved Work

​
Generation History Is Empty or Missing Items

Solutions:
  • Refresh the Captures/history view — entries are stored on PeachDesk’s servers and synced to your account.
  • Make sure you’re in the right workspace; history is per-workspace.
  • If you expected a specific take and it’s gone, check whether it was deleted or filtered by a search/date filter in the view.

​
Still Having Issues?

If you’re still experiencing problems:
  1. Check the docs: docs.peachdeskai.com
  2. Contact support: Reach out to the Bibha team from the PeachDesk dashboard. When you do, include:
    • What you were trying to do (tab and action — e.g. Generation, Effects, Voices)
    • The voice profile and model you selected
    • The exact error message, if any
    • Your browser and operating system
    • The approximate time the issue occurred (this helps us locate it in our logs)
Because Voice Studio is hosted, the Bibha team can investigate server-side issues directly from the details above — you don’t need to gather local logs or diagnostics.