---
title: "Audio splitter for speech-to-text limits: ffmpeg commands"
description: "Split long recordings to fit a transcription file limit: chunk count, overlap and ready ffmpeg commands for WhatsApp .ogg, .m4a and .opus to WAV or MP3."
url: https://broadrouter.com/tools/audio-splitter-ffmpeg/
lang: en
updated: 2026-11-15
publisher: BroadRouter (a BroadNet site); API by BroadNet Inference, https://inference.broadnet.ai
---

# Audio file splitter for speech-to-text limits

Updated 2026-11-15

**In short:** BroadNet's transcription endpoint accepts WAV or MP3 files up to 26,214,400 bytes. 16 kHz mono 16-bit WAV uses 32,000 bytes a second, so one file holds about 13 minutes; 64 kbps MP3 holds about 54. Enter a duration, or pick a file that stays on your device, for the chunk count and ready ffmpeg commands.

Free, no sign-up, runs in your browser: nothing you type is sent anywhere.

Audio and limitsAudio file (optional, read on your device only)Reads the name, size and duration in your browser. Nothing is uploaded.

DurationMinutes (90, 7.5) or h:mm:ss (1:30:00)Source fileWhatsApp .ogg (Opus)Phone .m4a (AAC).opus.mp3.wavOther (.webm, .aac, .flac)Input file nameOutput formatWAV, 16 kHz mono, 16-bit (lossless)MP3, 16 kHz mono, 64 kbpsMP3, 16 kHz mono, 32 kbps (smallest)MP3, 44.1 kHz mono, 128 kbpsSize limit, bytesSafety margin, %Overlap, secondsChunk length, secondsEmpty = longest that fits

Chunks

—

Chunk length

—

Est. size per chunk

—

Audio billed

—

## Size of the whole recording by format

| Format | Bytes per second | Est. file size | Longest chunk | Chunks needed |
| --- | --- | --- | --- | --- |

## Transcription cost

| Model | Per audio minute | This recording |
| --- | --- | --- |

[Get started](https://inference.broadnet.ai/account/?utm_source=broadrouter&utm_medium=referral&utm_campaign=tools-audio-splitter-ffmpeg&utm_content=tool) ·

New accounts get a $5 starting credit that covers both transcription models (and Gemma); DeepSeek, GLM and Qwen need a top-up. Accounts are managed on inference.broadnet.ai.

Sizes are estimates for constant-bitrate output plus up to 4 KB of headers, so keep a margin. Prices are BroadNet's published per-minute rates, synced 2026-11-14, applied to the audio duration including any overlap; BroadNet's /v1/models metadata (checked 2026-11-15) prices both transcription models per second of audio but does not say how part-seconds or per-request minimums are billed, so check your usage records. Before tax.

[Report a wrong price or number](mailto:info@broadnet.ai?subject=Wrong%20price%20or%20number%20on%20Audio%20file%20splitter%20for%20speech-to-text%20limits&body=Page%3A%20https%3A%2F%2Fbroadrouter.com%2Ftools%2Faudio-splitter-ffmpeg%2F%0A%0AWhat%20looks%20wrong%20(model%2C%20figure%20shown%2C%20correct%20figure%20and%20a%20source%20link%20if%20you%20have%20one)%3A%0A) (email to info@broadnet.ai, with this page's link and your inputs filled in).

## How it works

1.  Pick an audio file (it stays on your device: only its size and duration are read) or type the duration and choose the source type.
2.  Choose the output format: 16 kHz mono WAV, or MP3 at 32, 64 or 128 kbps. The size limit defaults to BroadNet's documented 26,214,400 bytes.
3.  Set a safety margin and an optional overlap in seconds. Leave chunk length empty to use the longest chunk that fits.
4.  Copy the ffmpeg commands: convert, split with -f segment, or cut overlapping chunks one by one. Read the transcription cost for both BroadNet models.

## How many minutes fit in one file?

File size = duration × bytes per second. Divide the limit by the bytes per second of your output format:

| Output | Bytes per second | Longest file under 26,214,400 bytes |
| --- | --- | --- |
| WAV, 16 kHz mono, 16-bit (lossless) | 32,000 | 13 min 39 s |
| MP3, 16 kHz mono, 64 kbps | 8,000 | 54 min 36 s |
| MP3, 16 kHz mono, 32 kbps (smallest) | 4,000 | 109 min 12 s |
| MP3, 44.1 kHz mono, 128 kbps | 16,000 | 27 min 18 s |
| WAV, 44.1 kHz stereo, 16-bit (not converted) | 176,400 | 2 min 28 s |

WAV bytes per second = sample rate × channels × 2 for 16-bit audio. MP3 bytes per second = bitrate in kbps × 1,000 ÷ 8.

## The ffmpeg commands, explained

```
ffmpeg -i recording.ogg -vn -ar 16000 -ac 1 -c:a pcm_s16le recording.wav
```

`-vn` drops any video or cover art, `-ar 16000` resamples to 16 kHz, `-ac 1` mixes to mono and `pcm_s16le` writes 16-bit WAV. For MP3, use `-c:a libmp3lame -b:a 64k` instead.

```
ffmpeg -i recording.ogg -vn -ar 16000 -ac 1 -c:a pcm_s16le -f segment -segment_time 778 -reset_timestamps 1 recording_%03d.wav
```

The segment muxer cuts the converted audio into consecutive files of `-segment_time` seconds, and `-reset_timestamps 1` starts each file at zero. FFmpeg's documentation warns that splitting may not be exact for video, where cuts fall on key frames; when converting to WAV or MP3 every audio frame can start a segment, so pieces land within a fraction of a second of the target.

For overlapping chunks, cut each one with `-ss` (start, in seconds) and `-t` (length). Putting `-ss` before `-i` seeks quickly; on compressed audio the start can be a few milliseconds off, which an overlap covers.

## Worked example

A 90-minute Arabic call converted to 16 kHz mono WAV is about 173 MB, over the 26,214,400-byte limit. With a 5% margin each chunk can be 778 seconds (12 min 58 s), so the call needs 7 chunks. At $0.0042 a minute on BroadNet Transcribe Arabic the 90 minutes cost $0.378. Converted to 64 kbps MP3 instead, the same call is about 43 MB and needs 2 files.

## Joining the transcripts

1.  Transcribe the chunks in order, with `language` set (BroadNet's docs say to always send it; use `ar` for Arabic).
2.  Join the texts with a space or a newline.
3.  If you used an overlap, remove the words repeated at each join: compare the last words of one chunk with the first words of the next.

The request itself is in the [Arabic speech-to-text quick start](https://broadrouter.com/guides/arabic-speech-to-text-api/).

## Questions

### What is the file size limit for BroadNet transcription?

26,214,400 bytes (25 MiB) per file, in WAV (audio/wav) or MP3 (audio/mpeg), according to BroadNet's API reference. A larger file is refused with HTTP 413, so split it first.

### How many minutes of audio fit in 25 MB?

It depends on the format, not the length. 16 kHz mono 16-bit WAV is 32,000 bytes a second, so 26,214,400 bytes hold about 13 min 39 s. 64 kbps MP3 is 8,000 bytes a second, about 54 min 36 s. Stereo or 44.1 kHz WAV holds far less: 44.1 kHz stereo 16-bit is 176,400 bytes a second, under 2.5 minutes.

### How do I convert a WhatsApp voice note for transcription?

WhatsApp voice notes are Opus audio in .ogg files. Run `ffmpeg -i voice-note.ogg -ar 16000 -ac 1 voice-note.wav`, the conversion BroadNet's docs show; the same works for .m4a and .opus. A voice note of a few minutes stays well under the limit. [Full guide](https://broadrouter.com/guides/transcribe-whatsapp-voice-notes/).

### Should chunks overlap?

A small overlap, such as 2 to 5 seconds, stops a word that falls on a cut from being lost in both chunks. It adds that much audio to every chunk after the first, so it adds billed minutes, and you remove the repeated words when you join the transcripts. With no overlap, split at silences if you can.

### Why does ffmpeg -f segment not overlap?

The segment muxer writes consecutive pieces and has no overlap option in the FFmpeg documentation. For overlapping chunks the tool writes one command per chunk with -ss (start) and -t (length), plus a shell loop.

### Have these ffmpeg commands been tested?

Yes. On 2026-11-15 we ran the commands this tool generates with FFmpeg 9.0.2 on a 30-minute Opus .ogg test file (a tone with noise, not real speech). We tried the 16 kHz WAV and 128 kbps MP3 outputs, with and without a 2-second overlap, and the 32 and 64 kbps MP3 single-file conversions. Every chunk stayed under 26,214,400 bytes (the largest was about 24.9 MB, inside the 5% margin), chunk lengths matched the plan to the second, and the predicted sizes were within about 1 KB of the real files. The ffprobe duration command was not part of that test.

### Is my audio file uploaded?

No. If you pick a file, the browser reads only its name, size and duration on your device; nothing is sent to BroadRouter or BroadNet. The commands run on your own machine.

### Is WAV or MP3 better for transcription?

WAV is lossless, so it avoids another round of compression on audio that is often already compressed, but it is large. MP3 at 64 kbps holds about four times as much audio per file. Both are accepted. If your source is a low-bitrate voice note, test both on a few files and compare the transcripts.

## Go deeper

-   [Transcribe WhatsApp voice notes with an API](https://broadrouter.com/guides/transcribe-whatsapp-voice-notes/)
-   [Arabic speech-to-text API: quick start](https://broadrouter.com/guides/arabic-speech-to-text-api/)
-   [Arabic transcription on BroadNet Inference](https://broadrouter.com/arabic-transcription/)
-   [Speech-to-text cost calculator](https://broadrouter.com/tools/speech-to-text-cost-calculator/)

## More free tools

-   [**LLM API cost calculator**Price a workload on every model at once, with cache hits, reasoning and a flat-plan check.](https://broadrouter.com/tools/llm-api-cost-calculator/)
-   [**Speech-to-text cost calculator**Hours, calls or voice notes to a monthly transcription bill, against your current provider.](https://broadrouter.com/tools/speech-to-text-cost-calculator/)
-   [**LLM token counter**Exact token counts with each model's real tokenizer, with a token view and cost.](https://broadrouter.com/tools/token-counter/)
-   [**Arabic token counter**Arabic vs English token counts per tokenizer: the "Arabic tax" in tokens and money.](https://broadrouter.com/tools/arabic-token-counter/)
-   [**Tokens to words converter**Tokens ↔ words ↔ characters ↔ pages, by language and content type.](https://broadrouter.com/tools/tokens-to-words/)
-   [**Context window calculator**Will it fit? Window used per model, room left for output, cost to send.](https://broadrouter.com/tools/context-window-calculator/)
-   [**Flat-rate vs per-token LLM cost calculator**Break-even between an unlimited plan and paying per token, per agent.](https://broadrouter.com/tools/flat-rate-vs-per-token-calculator/)
-   [**Prompt caching savings calculator**Bill with and without caching, by prefix share and hit rate.](https://broadrouter.com/tools/prompt-caching-calculator/)
-   [**Image token calculator**Tokens and cost per image under each model's documented rule. Drop an image; nothing is uploaded.](https://broadrouter.com/tools/image-token-calculator/)
-   [**Self-hosting vs API cost calculator**GPU rental against your current API and BroadNet per-token and flat-plan prices: cost per 1M tokens and break-even volume.](https://broadrouter.com/tools/self-hosting-vs-api-calculator/)
-   [**Embedding cost calculator**Index, update and query costs for a corpus at OpenAI, Google, Voyage, Mistral and Jina list prices.](https://broadrouter.com/tools/embedding-cost-calculator/)
-   [**LLM rate limit and concurrency calculator**Which limit binds first: RPM, TPM or concurrency, with Little's law.](https://broadrouter.com/tools/llm-rate-limit-calculator/)
-   [**JSON schema generator for LLM structured output**Sample JSON to strict schema, Pydantic, Zod and a ready request.](https://broadrouter.com/tools/json-schema-generator/)
-   [**OpenAI-compatible API request builder**One request, copied as curl, Python, Node or the OpenAI SDKs.](https://broadrouter.com/tools/api-request-builder/)
-   [**Agent run cost estimator**Context growth per step, total tokens and cost of one agent run, with a DeepSeek session comparison.](https://broadrouter.com/tools/agent-run-cost-estimator/)
-   [**OpenAI to BroadNet migration helper**Paste OpenAI SDK code; get base URL, key and model edits, a diff and the checks that remain.](https://broadrouter.com/tools/openai-to-broadnet-migration/)
-   [**Arabic text normalizer**Tashkeel, tatweel, alef, ya, ta marbuta, digits and invisible characters, with a matching Python function.](https://broadrouter.com/tools/arabic-text-normalizer/)
-   [**RAG chunking calculator**Chunks, embedding tokens, input per query, context check and monthly LLM cost for a RAG pipeline.](https://broadrouter.com/tools/rag-chunking-calculator/)
-   [**max\_tokens calculator for reasoning models**Answer length + reasoning budget + margin = a max\_tokens that does not cut the answer off.](https://broadrouter.com/tools/reasoning-token-budget/)
-   [**LLM batch job planner**Run time, binding limit, cost per model and the flat sessions needed to finish a big job by a deadline.](https://broadrouter.com/tools/llm-batch-job-planner/)
-   [**LLM JSON output validator**Parse errors by line and column, safe repairs, cut-off detection and JSON Schema checks for model output.](https://broadrouter.com/tools/llm-json-output-validator/)
-   [**Prompt token diff**Two prompt versions side by side: changed lines, exact token delta and the monthly cost difference.](https://broadrouter.com/tools/prompt-token-diff/)

Try it on your real workload. New accounts start with $5 of credit for Gemma and transcription. DeepSeek, GLM and Qwen3.8 need a top-up.

[Get started](https://inference.broadnet.ai/account/?utm_source=broadrouter&utm_medium=referral&utm_campaign=tools-audio-splitter-ffmpeg&utm_content=page-cta) ·

**Sources**

-   [BroadNet Inference machine-readable prices (pricing.json)](https://inference.broadnet.ai/pricing.json?utm_source=broadrouter&utm_medium=referral&utm_campaign=tools-audio-splitter-ffmpeg&utm_content=sources)
-   [BroadNet Inference API reference: transcription formats and 26,214,400-byte limit (checked 2026-11-15)](https://inference.broadnet.ai/docs?utm_source=broadrouter&utm_medium=referral&utm_campaign=tools-audio-splitter-ffmpeg&utm_content=sources)
-   [FFmpeg formats documentation: segment muxer (segment\_time, reset\_timestamps; checked 2026-11-15)](https://ffmpeg.org/ffmpeg-formats.html#segment)
-   [Meta WhatsApp Cloud API: audio messages (voice notes are OGG with the Opus codec)](https://developers.facebook.com/documentation/business-messaging/whatsapp/messages/audio-messages)
