Looking for a way to generate voiceovers for legacy broadcast equipment. Most TTS and AI voice APIs output modern formats like MP3 or AAC.
Our system requires:
* **Sample rate:** 44.1 kHz or 48 kHz
* **Bit depth:** 16-bit PCM
* **Format:** Uncompressed WAV (linear PCM) or sometimes specific broadcast WAV (BWF)
* **No codec issues**
Resemble AI's documentation mentions `WAV` export, but specifics are thin.
Has anyone used their API or web interface to successfully generate files that work with old hardware? Need to confirm:
* Is the WAV output truly uncompressed PCM?
* Can you control sample rate/bit depth?
* Are there any hidden headers or metadata that cause issues?
Tried a competitor's "WAV" output and it was actually a WAV container with ADPCM encoding, which failed. Don't want to waste time on another trial if the specs aren't met.
If you've done this, what was your exact export command or settings? Example API call would be key.
```bash
# Something like this?
curl -X POST https://app.resemble.ai/api/v1/projects/.../clips
-H "Authorization: Bearer YOUR_TOKEN"
-H "Content-Type: application/json"
-d '{"data":{"body":"Script text","format":"wav","sample_rate":44100}}'
```
Least privilege is not a suggestion.
Resemble's "WAV" is 22.05 kHz, 32-bit float PCM by default via API. Not your spec.
You'll need to pipe it through `ffmpeg` anyway. Don't overpay for their API calls for format conversion you'll do yourself. Most of these services are just wrapping cloud TTS APIs and marking up the cost.
```bash
curl ... [get audio from any cheap service]
ffmpeg -i input.wav -ar 48000 -acodec pcm_s16le output.wav
```
Check the headers with `file` or `mediainfo` before you push to broadcast gear.
show the math