PCM format admission: rates, channels and conversion are not device certification
This is a source/troubleshooting reference. Its samples do not establish that the store edition contains every described capability. For voice drafts into an existing workspace, see the verified 0.1.0 listing.
US listing verified; availability depends on your storefront. No Mac or self-signing for store installation. Bring your workspace address and account; the optional speech model is about 950 MB.
Within range is an admission prerequisite, not an all-accessory conversion guarantee
ColliePCMStreamConverter requires finite 8000–192000 Hz input, 1–8 channels and non-interleaved Float32 before attempting a 16 kHz mono output converter. Creation/conversion can still fail. This reviews exact admission, not a user rate control, hardware certification or universal audio importer.
Different from permission and route recovery
Permission checks prerequisites; recovery checks route events. Here why a native PCM format/block fails converter admission matters. No capture or format-coverage benchmark ran.
| Check | Verified scope | Manual review | Not established |
|---|---|---|---|
| Rate | Finite 8000…192000 Hz | Actual value/creation result | Every device in range tested |
| Channels | 1…8 | Actual channel descriptor | Speaker identification or track separation |
| Common format | pcmFormatFloat32 | Actual data, not suffix/label | Any integer PCM/compressed file directly accepted |
| Interleaving | Non-interleaved required | Actual buffer layout | Relabeling reinterprets bytes safely |
| Output | 16 kHz mono | Existing conversion succeeds | User 8 kHz bypass |
| Successive blocks | isStandard/rate/channels/format checks | Original sourceFormat match | Old converter always reusable |
| PCMChunk | Copies valid nonempty blocks across queues | Block/buffer lifetime | Privacy certification or recognition accuracy |
| Failure | unsupportedFormat/conversionFailed etc. | Format versus processing errors | Invented audio, cloud conversion or delivery |
Synthetic format review
Input rate is in range but data is not non-interleaved Float32. Result: still fails this converter’s prerequisites. A normal-looking rate is insufficient. Even passing all conditions needs successful creation/conversion.
Parameters are admission rules, not quality/accuracy promises. Do not falsify headers, force converters or skip destination/model guards. Error categories do not authenticate spoken content.
FAQ
Is 48000 Hz guaranteed usable?
No. Other conditions and actual creation/conversion must succeed.
Does 1–8 channels mean speaker separation?
No. It is input admission; output is mono, not diarization.
Can a Float32 label repair actual bytes?
No. Description/layout must match; labels cannot bypass checks.
Do matching channels justify reusing a converter?
Not alone. Rate, standard-format and other conditions also matter.
Is this hardware certification?
No device/noise/accuracy matrix was tested.
Does failure trigger cloud conversion or generated fill?
No fallback/fill is added. Keep errors/gaps within the original authorized process.
Reading and source
Permission/sample contract · Route recovery · Interruption gaps · Start · Privacy
