Text to Speech

Why do voices sound different on my phone versus my laptop?

Because voice selection and quality are provided by the operating system, not the browser or this tool. Apple's iOS and macOS ship high-quality neural voices that sound close to human; Android depends on Google's Play Services TTS and the specific Android version; Windows 10/11 includes modern Microsoft Neural voices but older Windows versions shipped lower-quality voices. The same text spoken by the 'default' voice on different devices will sound different because the default varies. If audio quality matters for your use case, specify a voice explicitly rather than relying on the default, and accept that some users on older hardware will hear lower-quality audio.

Can I download the generated audio as an MP3 or WAV file?

Not directly. The Web Speech API generates audio in real time through the OS audio pipeline and does not expose a 'save to file' method. To capture browser TTS as an audio file you would need to route the audio through a MediaRecorder API instance (which has limitations and inconsistent cross-browser support) or use a paid TTS service with a REST API that returns audio bytes (ElevenLabs, Google Cloud Text-to-Speech, Amazon Polly). For occasional personal use, screen-recording software can capture the audio, but for systematic generation of audio files for podcasts or videos, paid TTS APIs are the right tool.

What languages does this support?

Support depends on what voices your operating system has installed. Most modern OS installations cover the major European languages (English, Spanish, French, German, Italian, Portuguese), CJK (Chinese, Japanese, Korean), and common Asian languages (Arabic, Hindi, Vietnamese). Less-common languages (Swahili, Welsh, Icelandic) often require a one-time download of an additional language pack from the OS settings. Check your voice dropdown — if a language does not appear, visit your OS TTS settings to see if voices can be installed. Browser vendors occasionally bundle extra voices but OS-native voices generally sound better.

Why does the voice cut off partway through a long paragraph?

Chrome and some other browsers have an undocumented 200 to 300 character limit per SpeechSynthesisUtterance that causes longer texts to stop unexpectedly. Our tool works around this by splitting text into sentence-sized chunks and queuing them sequentially, so long-form content plays end to end. If you are using speechSynthesis directly in your own code and seeing this issue, implement the same chunking: split on terminal punctuation, create one SpeechSynthesisUtterance per chunk, and queue them all before calling speak() on the first.

Is this suitable for creating commercial video voiceover or audiobooks?

Generally no. Browser TTS is a prototype and accessibility tool, not a production voiceover service. For commercial content, the audio quality varies too much across listener devices, you cannot download the audio as files, and the licensing of system voices for redistribution is restrictive (Apple and Microsoft voices are not licensed for commercial distribution of the generated audio). For production voiceover, paid services like ElevenLabs (consistent high-quality neural voices, commercial licensing clear), Google Cloud Text-to-Speech, Amazon Polly, or Azure Cognitive Services Speech provide consistent output you can use in ads, videos, and audiobooks without legal or quality concerns.