Every TTS platform asks you to pick an output format. WAV, MP3, and OGG Opus each have real tradeoffs — and the right choice depends on what you're building, not just which one sounds best in a side-by-side test. Here's a practical breakdown.

The Three Formats at a Glance

WAV (Waveform Audio)

• Compression: None (uncompressed PCM)
• Quality: Lossless — exactly what the TTS engine produced
• File size: Large. A 60-second clip at 44.1 kHz stereo is roughly 10 MB
• Compatibility: Universal — every audio tool, browser, and OS handles WAV

MP3 (MPEG Audio Layer III)

• Compression: Lossy
• Quality: Good to excellent depending on bitrate (128–320 kbps typical)
• File size: Small. Roughly 1/10th the size of WAV at 128 kbps
• Compatibility: Universal — the most widely supported compressed audio format

OGG Opus

• Compression: Lossy
• Quality: Excellent — generally better than MP3 at the same bitrate
• File size: Small. Comparable to or smaller than MP3 at equivalent quality
• Compatibility: Good — supported in modern browsers and most media players, but not universal in legacy systems

When to Use WAV

WAV is the right default when quality preservation matters more than file size. Specific scenarios:

• Post-production editing — if you're importing TTS audio into a DAW (Audacity, Adobe Audition, Logic) for mixing, effects, or mastering, start with WAV. Every re-encode of a lossy format degrades quality slightly.
• Archival — if you want a master copy that can be re-encoded to any format later without generation loss.
• Short clips — for notifications, UI sounds, or short prompts, the file size difference is negligible and WAV avoids any compression artifacts.

WAV is the default output format in most TTS platforms, including AI TTS Microservice, for good reason: it's the safest starting point.

When to Use MP3

MP3 makes sense when you need small files and broad compatibility:

• Web delivery — embedding audio on a webpage, email, or CMS where bandwidth matters.
• Podcast distribution — most podcast hosts and RSS feeds expect MP3.
• Mobile apps — smaller downloads and less storage consumption on device.
• Legacy systems — if your pipeline or downstream consumer only accepts MP3.

At 128 kbps, MP3 is transparent for speech — meaning most listeners can't distinguish it from the uncompressed original. For voice-only content (no music, no complex audio), 128 kbps is a solid default. Go higher (192–320 kbps) only if the audio will be mixed with music or if your audience uses high-end monitoring equipment.

When to Use OGG Opus

OGG Opus is the technically superior lossy format, but compatibility is the limiting factor:

• Modern web apps — all current browsers (Chrome, Firefox, Safari, Edge) support Opus natively. If your audience is on modern browsers, Opus gives you better quality at smaller file sizes than MP3.
• Real-time streaming — Opus was designed for low-latency audio. If you're streaming TTS output in a chat interface or voice assistant, Opus is the natural choice.
• Bandwidth-constrained delivery — Opus at 64 kbps sounds comparable to MP3 at 128 kbps for speech content. That's a meaningful saving at scale.

Avoid OGG Opus when your downstream system doesn't support it — some older media players, embedded devices, and enterprise audio pipelines still expect MP3 or WAV only.

Bitrate and Sample Rate: What Actually Matters for Speech

TTS audio is simpler than music — it's mostly a single voice in a narrow frequency range. That means you can use lower bitrates and sample rates without audible loss:

• Sample rate: 24 kHz is sufficient for clear speech. 44.1 kHz or 48 kHz adds no perceptible benefit for voice-only content but doubles the file size. Use higher sample rates only if you're mixing with music or need to match an existing production pipeline.
• MP3 bitrate: 128 kbps is transparent for speech. 64 kbps is acceptable for low-bandwidth scenarios. Below 64 kbps, artifacts become noticeable.
• Opus bitrate: 64 kbps is excellent for speech. 32 kbps is usable for voice-only content in constrained environments.

Some TTS platforms let you configure both bitrate and sample rate per request. If yours does, match the settings to your delivery channel rather than defaulting to maximum quality.

A Practical Decision Framework

• Editing the audio afterward? → WAV
• Delivering to a broad audience or legacy system? → MP3 at 128 kbps
• Modern web app with bandwidth sensitivity? → OGG Opus at 64 kbps
• Short UI sounds or notifications? → WAV (size is negligible)
• Podcast or RSS feed? → MP3
• Not sure? → WAV, then convert later as needed

---

Try it: Generate audio in WAV, MP3, or OGG Opus and compare the output yourself. All three formats are available on every voice.